new_stack_ai_skills_tracker_10jul24.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": [],
      "authorship_tag": "ABX9TyPQGN3TPPYq6V4RJyYcHiZT",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/bradshimmin/llm_experiments/blob/main/new_stack_ai_skills_tracker_10jul24.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# GenAI Technology Stack Tracker"
      ],
      "metadata": {
        "id": "RB8wRf8Rqj6J"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "---\n",
        "## Install and import libraries"
      ],
      "metadata": {
        "id": "Y_qwNPoPqqyg"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Install and import necessary libraries\n",
        "import pandas as pd\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "import seaborn as sns\n",
        "from IPython.display import display, Markdown\n",
        "import os\n",
        "import json\n",
        "import re\n",
        "from pprint import pprint\n"
      ],
      "metadata": {
        "id": "oAM_E7exscTN"
      },
      "execution_count": 1,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# Mount Google Drive\n",
        "from google.colab import drive\n",
        "drive.mount('/content/gdrive')\n",
        "\n",
        "# Get secret keys\n",
        "from google.colab import userdata\n"
      ],
      "metadata": {
        "id": "YpDHdpNg-kY9",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "4ac7b42e-d12a-41e1-a213-fc81a73b22b0"
      },
      "execution_count": 2,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Mounted at /content/gdrive\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 3,
      "metadata": {
        "id": "dSCtobQF-iPp",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "d56367be-edd3-4206-d5f6-47e8b1d972be"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m328.5/328.5 kB\u001b[0m \u001b[31m2.5 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m75.6/75.6 kB\u001b[0m \u001b[31m6.1 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m77.9/77.9 kB\u001b[0m \u001b[31m8.6 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m58.3/58.3 kB\u001b[0m \u001b[31m6.1 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m865.4/865.4 kB\u001b[0m \u001b[31m5.0 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m318.9/318.9 kB\u001b[0m \u001b[31m17.9 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[?25h"
          ]
        }
      ],
      "source": [
        "# Install OpenAI and Anthropic libraries\n",
        "!pip install -qU openai\n",
        "from openai import OpenAI\n",
        "\n",
        "# set up Anthropic\n",
        "!pip install -qU anthropic\n",
        "from anthropic import Anthropic\n"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Set up OpenAI\n",
        "openai_model=\"gpt-3.5-turbo-instruct\"\n",
        "openai_client = OpenAI(api_key=userdata.get('OpenAI_key'))\n",
        "\n",
        "# set up Anthropic\n",
        "anthropic_model = \"claude-3-haiku-20240307\"\n",
        "anthropic_client = Anthropic(api_key=userdata.get('Anthropic_key'))"
      ],
      "metadata": {
        "id": "_OkAFNlL4pTY"
      },
      "execution_count": 4,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "---\n",
        "## Get some data\n"
      ],
      "metadata": {
        "id": "_TCjG492qg6P"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Load DF\n",
        "df = pd.read_pickle('/content/gdrive/MyDrive/Colab Notebooks/Datasets/ai_skills_tracker/ai_skills_2q24/AI Skills Full 2Q24.pkl')\n",
        "print(\"all done\")\n"
      ],
      "metadata": {
        "id": "_702AG-a-kWK",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "12be4882-61ce-4803-ed7d-63e251f625f2"
      },
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "all done\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "df.shape"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "0XSB5nyUbxep",
        "outputId": "07ffaab0-e045-4ad8-e422-09ea85eedaa8"
      },
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(205517, 30)"
            ]
          },
          "metadata": {},
          "execution_count": 6
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "df.info()"
      ],
      "metadata": {
        "id": "5X9xZQx_-kRM",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "78257841-ca7d-4bd8-bb46-3fdce6070118"
      },
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "<class 'pandas.core.frame.DataFrame'>\n",
            "Index: 205517 entries, 0 to 466806\n",
            "Data columns (total 30 columns):\n",
            " #   Column                    Non-Null Count   Dtype  \n",
            "---  ------                    --------------   -----  \n",
            " 0   RecordID                  205517 non-null  int64  \n",
            " 1   Job title                 205517 non-null  object \n",
            " 2   Full job description      205517 non-null  object \n",
            " 3   Date collected            160769 non-null  object \n",
            " 4   Date posted               205517 non-null  object \n",
            " 5   Week                      205517 non-null  int64  \n",
            " 6   Year                      205517 non-null  int64  \n",
            " 7   Month                     205517 non-null  int64  \n",
            " 8   Quarter                   205517 non-null  object \n",
            " 9   Hiring entity             205517 non-null  object \n",
            " 10  Entity size               194224 non-null  object \n",
            " 11  Home office hybrid work   205517 non-null  object \n",
            " 12  Full or part time status  202508 non-null  object \n",
            " 13  Number of applicants      83853 non-null   float64\n",
            " 14  Longitude                 205517 non-null  float64\n",
            " 15  Latitude                  205517 non-null  float64\n",
            " 16  City                      202294 non-null  object \n",
            " 17  Territory                 205497 non-null  object \n",
            " 18  Region                    205483 non-null  object \n",
            " 19  Experience level          203906 non-null  object \n",
            " 20  Education level           192103 non-null  object \n",
            " 21  Salary start              109670 non-null  float64\n",
            " 22  Salary end                108908 non-null  float64\n",
            " 23  Main industry             198790 non-null  object \n",
            " 24  Job title category        205517 non-null  object \n",
            " 25  AI technologies           205517 non-null  object \n",
            " 26  AI skills                 205517 non-null  object \n",
            " 27  Source name               205517 non-null  object \n",
            " 28  Source URL                205517 non-null  object \n",
            " 29  File Date                 205517 non-null  object \n",
            "dtypes: float64(5), int64(4), object(21)\n",
            "memory usage: 48.6+ MB\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "df.describe()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 300
        },
        "id": "ybCY4TS6fZIu",
        "outputId": "eecd8512-2c3e-4572-d740-1a62bafef8b4"
      },
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "            RecordID           Week           Year          Month  \\\n",
              "count  205517.000000  205517.000000  205517.000000  205517.000000   \n",
              "mean   153375.824073      25.749378    2022.844392       6.355158   \n",
              "std    133859.621951      14.756254       0.722429       3.409576   \n",
              "min         1.000000       1.000000    2022.000000       1.000000   \n",
              "25%     52246.000000      14.000000    2022.000000       4.000000   \n",
              "50%    104686.000000      24.000000    2023.000000       6.000000   \n",
              "75%    274411.000000      38.000000    2023.000000       9.000000   \n",
              "max    466806.000000      52.000000    2024.000000      12.000000   \n",
              "\n",
              "       Number of applicants      Longitude       Latitude   Salary start  \\\n",
              "count          83853.000000  205517.000000  205517.000000  109670.000000   \n",
              "mean              32.844967     -55.994817      36.510116   88063.251518   \n",
              "std               42.365210      69.150189      14.424002   22601.000215   \n",
              "min                0.000000    -158.081787     -54.808030       0.000000   \n",
              "25%                4.000000     -99.414619      34.053570   81000.000000   \n",
              "50%               15.000000     -79.644378      39.398703   88602.000000   \n",
              "75%               45.000000      -1.550268      42.358662   95000.000000   \n",
              "max              200.000000     179.259100      70.027908  500000.000000   \n",
              "\n",
              "          Salary end  \n",
              "count  108908.000000  \n",
              "mean   131141.707937  \n",
              "std     33389.021447  \n",
              "min         0.000000  \n",
              "25%    112863.500000  \n",
              "50%    127559.000000  \n",
              "75%    147590.000000  \n",
              "max    993404.000000  "
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-1952c4c1-6968-41be-ad39-691c2fdfc590\" class=\"colab-df-container\">\n",
              "    <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>RecordID</th>\n",
              "      <th>Week</th>\n",
              "      <th>Year</th>\n",
              "      <th>Month</th>\n",
              "      <th>Number of applicants</th>\n",
              "      <th>Longitude</th>\n",
              "      <th>Latitude</th>\n",
              "      <th>Salary start</th>\n",
              "      <th>Salary end</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>count</th>\n",
              "      <td>205517.000000</td>\n",
              "      <td>205517.000000</td>\n",
              "      <td>205517.000000</td>\n",
              "      <td>205517.000000</td>\n",
              "      <td>83853.000000</td>\n",
              "      <td>205517.000000</td>\n",
              "      <td>205517.000000</td>\n",
              "      <td>109670.000000</td>\n",
              "      <td>108908.000000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>mean</th>\n",
              "      <td>153375.824073</td>\n",
              "      <td>25.749378</td>\n",
              "      <td>2022.844392</td>\n",
              "      <td>6.355158</td>\n",
              "      <td>32.844967</td>\n",
              "      <td>-55.994817</td>\n",
              "      <td>36.510116</td>\n",
              "      <td>88063.251518</td>\n",
              "      <td>131141.707937</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>std</th>\n",
              "      <td>133859.621951</td>\n",
              "      <td>14.756254</td>\n",
              "      <td>0.722429</td>\n",
              "      <td>3.409576</td>\n",
              "      <td>42.365210</td>\n",
              "      <td>69.150189</td>\n",
              "      <td>14.424002</td>\n",
              "      <td>22601.000215</td>\n",
              "      <td>33389.021447</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>min</th>\n",
              "      <td>1.000000</td>\n",
              "      <td>1.000000</td>\n",
              "      <td>2022.000000</td>\n",
              "      <td>1.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>-158.081787</td>\n",
              "      <td>-54.808030</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>25%</th>\n",
              "      <td>52246.000000</td>\n",
              "      <td>14.000000</td>\n",
              "      <td>2022.000000</td>\n",
              "      <td>4.000000</td>\n",
              "      <td>4.000000</td>\n",
              "      <td>-99.414619</td>\n",
              "      <td>34.053570</td>\n",
              "      <td>81000.000000</td>\n",
              "      <td>112863.500000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>50%</th>\n",
              "      <td>104686.000000</td>\n",
              "      <td>24.000000</td>\n",
              "      <td>2023.000000</td>\n",
              "      <td>6.000000</td>\n",
              "      <td>15.000000</td>\n",
              "      <td>-79.644378</td>\n",
              "      <td>39.398703</td>\n",
              "      <td>88602.000000</td>\n",
              "      <td>127559.000000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>75%</th>\n",
              "      <td>274411.000000</td>\n",
              "      <td>38.000000</td>\n",
              "      <td>2023.000000</td>\n",
              "      <td>9.000000</td>\n",
              "      <td>45.000000</td>\n",
              "      <td>-1.550268</td>\n",
              "      <td>42.358662</td>\n",
              "      <td>95000.000000</td>\n",
              "      <td>147590.000000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>max</th>\n",
              "      <td>466806.000000</td>\n",
              "      <td>52.000000</td>\n",
              "      <td>2024.000000</td>\n",
              "      <td>12.000000</td>\n",
              "      <td>200.000000</td>\n",
              "      <td>179.259100</td>\n",
              "      <td>70.027908</td>\n",
              "      <td>500000.000000</td>\n",
              "      <td>993404.000000</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "    <div class=\"colab-df-buttons\">\n",
              "\n",
              "  <div class=\"colab-df-container\">\n",
              "    <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-1952c4c1-6968-41be-ad39-691c2fdfc590')\"\n",
              "            title=\"Convert this dataframe to an interactive table.\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\" viewBox=\"0 -960 960 960\">\n",
              "    <path d=\"M120-120v-720h720v720H120Zm60-500h600v-160H180v160Zm220 220h160v-160H400v160Zm0 220h160v-160H400v160ZM180-400h160v-160H180v160Zm440 0h160v-160H620v160ZM180-180h160v-160H180v160Zm440 0h160v-160H620v160Z\"/>\n",
              "  </svg>\n",
              "    </button>\n",
              "\n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    .colab-df-buttons div {\n",
              "      margin-bottom: 4px;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "    <script>\n",
              "      const buttonEl =\n",
              "        document.querySelector('#df-1952c4c1-6968-41be-ad39-691c2fdfc590 button.colab-df-convert');\n",
              "      buttonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "      async function convertToInteractive(key) {\n",
              "        const element = document.querySelector('#df-1952c4c1-6968-41be-ad39-691c2fdfc590');\n",
              "        const dataTable =\n",
              "          await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                    [key], {});\n",
              "        if (!dataTable) return;\n",
              "\n",
              "        const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "          '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "          + ' to learn more about interactive tables.';\n",
              "        element.innerHTML = '';\n",
              "        dataTable['output_type'] = 'display_data';\n",
              "        await google.colab.output.renderOutput(dataTable, element);\n",
              "        const docLink = document.createElement('div');\n",
              "        docLink.innerHTML = docLinkHtml;\n",
              "        element.appendChild(docLink);\n",
              "      }\n",
              "    </script>\n",
              "  </div>\n",
              "\n",
              "\n",
              "<div id=\"df-52740704-656b-44af-889a-6b946e95c244\">\n",
              "  <button class=\"colab-df-quickchart\" onclick=\"quickchart('df-52740704-656b-44af-889a-6b946e95c244')\"\n",
              "            title=\"Suggest charts\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "<svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "     width=\"24px\">\n",
              "    <g>\n",
              "        <path d=\"M19 3H5c-1.1 0-2 .9-2 2v14c0 1.1.9 2 2 2h14c1.1 0 2-.9 2-2V5c0-1.1-.9-2-2-2zM9 17H7v-7h2v7zm4 0h-2V7h2v10zm4 0h-2v-4h2v4z\"/>\n",
              "    </g>\n",
              "</svg>\n",
              "  </button>\n",
              "\n",
              "<style>\n",
              "  .colab-df-quickchart {\n",
              "      --bg-color: #E8F0FE;\n",
              "      --fill-color: #1967D2;\n",
              "      --hover-bg-color: #E2EBFA;\n",
              "      --hover-fill-color: #174EA6;\n",
              "      --disabled-fill-color: #AAA;\n",
              "      --disabled-bg-color: #DDD;\n",
              "  }\n",
              "\n",
              "  [theme=dark] .colab-df-quickchart {\n",
              "      --bg-color: #3B4455;\n",
              "      --fill-color: #D2E3FC;\n",
              "      --hover-bg-color: #434B5C;\n",
              "      --hover-fill-color: #FFFFFF;\n",
              "      --disabled-bg-color: #3B4455;\n",
              "      --disabled-fill-color: #666;\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart {\n",
              "    background-color: var(--bg-color);\n",
              "    border: none;\n",
              "    border-radius: 50%;\n",
              "    cursor: pointer;\n",
              "    display: none;\n",
              "    fill: var(--fill-color);\n",
              "    height: 32px;\n",
              "    padding: 0;\n",
              "    width: 32px;\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart:hover {\n",
              "    background-color: var(--hover-bg-color);\n",
              "    box-shadow: 0 1px 2px rgba(60, 64, 67, 0.3), 0 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "    fill: var(--button-hover-fill-color);\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart-complete:disabled,\n",
              "  .colab-df-quickchart-complete:disabled:hover {\n",
              "    background-color: var(--disabled-bg-color);\n",
              "    fill: var(--disabled-fill-color);\n",
              "    box-shadow: none;\n",
              "  }\n",
              "\n",
              "  .colab-df-spinner {\n",
              "    border: 2px solid var(--fill-color);\n",
              "    border-color: transparent;\n",
              "    border-bottom-color: var(--fill-color);\n",
              "    animation:\n",
              "      spin 1s steps(1) infinite;\n",
              "  }\n",
              "\n",
              "  @keyframes spin {\n",
              "    0% {\n",
              "      border-color: transparent;\n",
              "      border-bottom-color: var(--fill-color);\n",
              "      border-left-color: var(--fill-color);\n",
              "    }\n",
              "    20% {\n",
              "      border-color: transparent;\n",
              "      border-left-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "    }\n",
              "    30% {\n",
              "      border-color: transparent;\n",
              "      border-left-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "      border-right-color: var(--fill-color);\n",
              "    }\n",
              "    40% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "    }\n",
              "    60% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "    }\n",
              "    80% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "      border-bottom-color: var(--fill-color);\n",
              "    }\n",
              "    90% {\n",
              "      border-color: transparent;\n",
              "      border-bottom-color: var(--fill-color);\n",
              "    }\n",
              "  }\n",
              "</style>\n",
              "\n",
              "  <script>\n",
              "    async function quickchart(key) {\n",
              "      const quickchartButtonEl =\n",
              "        document.querySelector('#' + key + ' button');\n",
              "      quickchartButtonEl.disabled = true;  // To prevent multiple clicks.\n",
              "      quickchartButtonEl.classList.add('colab-df-spinner');\n",
              "      try {\n",
              "        const charts = await google.colab.kernel.invokeFunction(\n",
              "            'suggestCharts', [key], {});\n",
              "      } catch (error) {\n",
              "        console.error('Error during call to suggestCharts:', error);\n",
              "      }\n",
              "      quickchartButtonEl.classList.remove('colab-df-spinner');\n",
              "      quickchartButtonEl.classList.add('colab-df-quickchart-complete');\n",
              "    }\n",
              "    (() => {\n",
              "      let quickchartButtonEl =\n",
              "        document.querySelector('#df-52740704-656b-44af-889a-6b946e95c244 button');\n",
              "      quickchartButtonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "    })();\n",
              "  </script>\n",
              "</div>\n",
              "\n",
              "    </div>\n",
              "  </div>\n"
            ],
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "dataframe",
              "summary": "{\n  \"name\": \"df\",\n  \"rows\": 8,\n  \"fields\": [\n    {\n      \"column\": \"RecordID\",\n      \"properties\": {\n        \"dtype\": \"number\",\n        \"std\": 145774.59065452684,\n        \"min\": 1.0,\n        \"max\": 466806.0,\n        \"num_unique_values\": 8,\n        \"samples\": [\n          153375.82407294773,\n          104686.0,\n          205517.0\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    },\n    {\n      \"column\": \"Week\",\n      \"properties\": {\n        \"dtype\": \"number\",\n        \"std\": 72652.67251585863,\n        \"min\": 1.0,\n        \"max\": 205517.0,\n        \"num_unique_values\": 8,\n        \"samples\": [\n          25.749378396920935,\n          24.0,\n          205517.0\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    },\n    {\n      \"column\": \"Year\",\n      \"properties\": {\n        \"dtype\": \"number\",\n        \"std\": 72051.66696164048,\n        \"min\": 0.7224293955808271,\n        \"max\": 205517.0,\n        \"num_unique_values\": 6,\n        \"samples\": [\n          205517.0,\n          2022.844392434689,\n          2024.0\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    },\n    {\n      \"column\": \"Month\",\n      \"properties\": {\n        \"dtype\": \"number\",\n        \"std\": 72659.12281642767,\n        \"min\": 1.0,\n        \"max\": 205517.0,\n        \"num_unique_values\": 8,\n        \"samples\": [\n          6.355157967467411,\n          6.0,\n          205517.0\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    },\n    {\n      \"column\": \"Number of applicants\",\n      \"properties\": {\n        \"dtype\": \"number\",\n        \"std\": 29629.448952947623,\n        \"min\": 0.0,\n        \"max\": 83853.0,\n        \"num_unique_values\": 8,\n        \"samples\": [\n          32.844966787115546,\n          15.0,\n          83853.0\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    },\n    {\n      \"column\": \"Longitude\",\n      \"properties\": {\n        \"dtype\": \"number\",\n        \"std\": 72668.69713587883,\n        \"min\": -158.081787109375,\n        \"max\": 205517.0,\n        \"num_unique_values\": 8,\n        \"samples\": [\n          -55.99481678145987,\n          -79.644378,\n          205517.0\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    },\n    {\n      \"column\": \"Latitude\",\n      \"properties\": {\n        \"dtype\": \"number\",\n        \"std\": 72652.05062715481,\n        \"min\": -54.80803,\n        \"max\": 205517.0,\n        \"num_unique_values\": 8,\n        \"samples\": [\n          36.51011553638645,\n          39.39870315600007,\n          205517.0\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    },\n    {\n      \"column\": \"Salary start\",\n      \"properties\": {\n        \"dtype\": \"number\",\n        \"std\": 156965.57278834964,\n        \"min\": 0.0,\n        \"max\": 500000.0,\n        \"num_unique_values\": 8,\n        \"samples\": [\n          88063.25151819094,\n          88602.0,\n          109670.0\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    },\n    {\n      \"column\": \"Salary end\",\n      \"properties\": {\n        \"dtype\": \"number\",\n        \"std\": 321935.99218071543,\n        \"min\": 0.0,\n        \"max\": 993404.0,\n        \"num_unique_values\": 8,\n        \"samples\": [\n          131141.70793697433,\n          127559.0,\n          108908.0\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    }\n  ]\n}"
            }
          },
          "metadata": {},
          "execution_count": 8
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Convert \"Date posted\" to datetime format\n",
        "df['Date posted'] = pd.to_datetime(df['Date posted'])\n",
        "\n",
        "# Create a new dataframe with only entries in 2024\n",
        "df_2024 = df[df['Date posted'].dt.year == 2024]\n",
        "\n",
        "df_2024.shape\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Iw2WZYwp2P_N",
        "outputId": "5970f04f-f4f0-4048-a67e-e4fa2a3837b2"
      },
      "execution_count": 9,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(40128, 30)"
            ]
          },
          "metadata": {},
          "execution_count": 9
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# pull out entries that specify GenAI tech\n",
        "keywords = ['GenAI', 'transformer', 'openai', 'gpt', 'generative AI']\n",
        "\n",
        "df_genai = df_2024[df_2024['Full job description'].str.contains('|'.join(keywords))]\n",
        "\n",
        "df_genai.shape\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "a4NIAeAT3Evq",
        "outputId": "34252b3d-f16d-4f1e-ce14-18230a32d883"
      },
      "execution_count": 10,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(2880, 30)"
            ]
          },
          "metadata": {},
          "execution_count": 10
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Create a test sample from this sete\n",
        "openai_small_sample_df = df_genai[['RecordID', 'Full job description']].sample(n=2)\n",
        "anthropic_small_sample_df = openai_small_sample_df.copy()"
      ],
      "metadata": {
        "id": "NGo-QFlS9tVZ"
      },
      "execution_count": 11,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# Check sample\n",
        "openai_small_sample_df"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 125
        },
        "id": "UGA8kV33r2OH",
        "outputId": "c50be5e3-24a7-4f05-b25c-e4852a320cf0"
      },
      "execution_count": 12,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "        RecordID                               Full job description\n",
              "274043    274046  Please return soon to complete your dream job ...\n",
              "454763    454763  FEQ225R94\\n\\nThis role can be remote, targetin..."
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-1ad6e188-a83b-406f-8e27-5a64c680c1a1\" class=\"colab-df-container\">\n",
              "    <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>RecordID</th>\n",
              "      <th>Full job description</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>274043</th>\n",
              "      <td>274046</td>\n",
              "      <td>Please return soon to complete your dream job ...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>454763</th>\n",
              "      <td>454763</td>\n",
              "      <td>FEQ225R94\\n\\nThis role can be remote, targetin...</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "    <div class=\"colab-df-buttons\">\n",
              "\n",
              "  <div class=\"colab-df-container\">\n",
              "    <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-1ad6e188-a83b-406f-8e27-5a64c680c1a1')\"\n",
              "            title=\"Convert this dataframe to an interactive table.\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\" viewBox=\"0 -960 960 960\">\n",
              "    <path d=\"M120-120v-720h720v720H120Zm60-500h600v-160H180v160Zm220 220h160v-160H400v160Zm0 220h160v-160H400v160ZM180-400h160v-160H180v160Zm440 0h160v-160H620v160ZM180-180h160v-160H180v160Zm440 0h160v-160H620v160Z\"/>\n",
              "  </svg>\n",
              "    </button>\n",
              "\n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    .colab-df-buttons div {\n",
              "      margin-bottom: 4px;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "    <script>\n",
              "      const buttonEl =\n",
              "        document.querySelector('#df-1ad6e188-a83b-406f-8e27-5a64c680c1a1 button.colab-df-convert');\n",
              "      buttonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "      async function convertToInteractive(key) {\n",
              "        const element = document.querySelector('#df-1ad6e188-a83b-406f-8e27-5a64c680c1a1');\n",
              "        const dataTable =\n",
              "          await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                    [key], {});\n",
              "        if (!dataTable) return;\n",
              "\n",
              "        const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "          '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "          + ' to learn more about interactive tables.';\n",
              "        element.innerHTML = '';\n",
              "        dataTable['output_type'] = 'display_data';\n",
              "        await google.colab.output.renderOutput(dataTable, element);\n",
              "        const docLink = document.createElement('div');\n",
              "        docLink.innerHTML = docLinkHtml;\n",
              "        element.appendChild(docLink);\n",
              "      }\n",
              "    </script>\n",
              "  </div>\n",
              "\n",
              "\n",
              "<div id=\"df-0cc0d13c-68b6-408f-bc19-cbf838a484a9\">\n",
              "  <button class=\"colab-df-quickchart\" onclick=\"quickchart('df-0cc0d13c-68b6-408f-bc19-cbf838a484a9')\"\n",
              "            title=\"Suggest charts\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "<svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "     width=\"24px\">\n",
              "    <g>\n",
              "        <path d=\"M19 3H5c-1.1 0-2 .9-2 2v14c0 1.1.9 2 2 2h14c1.1 0 2-.9 2-2V5c0-1.1-.9-2-2-2zM9 17H7v-7h2v7zm4 0h-2V7h2v10zm4 0h-2v-4h2v4z\"/>\n",
              "    </g>\n",
              "</svg>\n",
              "  </button>\n",
              "\n",
              "<style>\n",
              "  .colab-df-quickchart {\n",
              "      --bg-color: #E8F0FE;\n",
              "      --fill-color: #1967D2;\n",
              "      --hover-bg-color: #E2EBFA;\n",
              "      --hover-fill-color: #174EA6;\n",
              "      --disabled-fill-color: #AAA;\n",
              "      --disabled-bg-color: #DDD;\n",
              "  }\n",
              "\n",
              "  [theme=dark] .colab-df-quickchart {\n",
              "      --bg-color: #3B4455;\n",
              "      --fill-color: #D2E3FC;\n",
              "      --hover-bg-color: #434B5C;\n",
              "      --hover-fill-color: #FFFFFF;\n",
              "      --disabled-bg-color: #3B4455;\n",
              "      --disabled-fill-color: #666;\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart {\n",
              "    background-color: var(--bg-color);\n",
              "    border: none;\n",
              "    border-radius: 50%;\n",
              "    cursor: pointer;\n",
              "    display: none;\n",
              "    fill: var(--fill-color);\n",
              "    height: 32px;\n",
              "    padding: 0;\n",
              "    width: 32px;\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart:hover {\n",
              "    background-color: var(--hover-bg-color);\n",
              "    box-shadow: 0 1px 2px rgba(60, 64, 67, 0.3), 0 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "    fill: var(--button-hover-fill-color);\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart-complete:disabled,\n",
              "  .colab-df-quickchart-complete:disabled:hover {\n",
              "    background-color: var(--disabled-bg-color);\n",
              "    fill: var(--disabled-fill-color);\n",
              "    box-shadow: none;\n",
              "  }\n",
              "\n",
              "  .colab-df-spinner {\n",
              "    border: 2px solid var(--fill-color);\n",
              "    border-color: transparent;\n",
              "    border-bottom-color: var(--fill-color);\n",
              "    animation:\n",
              "      spin 1s steps(1) infinite;\n",
              "  }\n",
              "\n",
              "  @keyframes spin {\n",
              "    0% {\n",
              "      border-color: transparent;\n",
              "      border-bottom-color: var(--fill-color);\n",
              "      border-left-color: var(--fill-color);\n",
              "    }\n",
              "    20% {\n",
              "      border-color: transparent;\n",
              "      border-left-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "    }\n",
              "    30% {\n",
              "      border-color: transparent;\n",
              "      border-left-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "      border-right-color: var(--fill-color);\n",
              "    }\n",
              "    40% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "    }\n",
              "    60% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "    }\n",
              "    80% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "      border-bottom-color: var(--fill-color);\n",
              "    }\n",
              "    90% {\n",
              "      border-color: transparent;\n",
              "      border-bottom-color: var(--fill-color);\n",
              "    }\n",
              "  }\n",
              "</style>\n",
              "\n",
              "  <script>\n",
              "    async function quickchart(key) {\n",
              "      const quickchartButtonEl =\n",
              "        document.querySelector('#' + key + ' button');\n",
              "      quickchartButtonEl.disabled = true;  // To prevent multiple clicks.\n",
              "      quickchartButtonEl.classList.add('colab-df-spinner');\n",
              "      try {\n",
              "        const charts = await google.colab.kernel.invokeFunction(\n",
              "            'suggestCharts', [key], {});\n",
              "      } catch (error) {\n",
              "        console.error('Error during call to suggestCharts:', error);\n",
              "      }\n",
              "      quickchartButtonEl.classList.remove('colab-df-spinner');\n",
              "      quickchartButtonEl.classList.add('colab-df-quickchart-complete');\n",
              "    }\n",
              "    (() => {\n",
              "      let quickchartButtonEl =\n",
              "        document.querySelector('#df-0cc0d13c-68b6-408f-bc19-cbf838a484a9 button');\n",
              "      quickchartButtonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "    })();\n",
              "  </script>\n",
              "</div>\n",
              "\n",
              "  <div id=\"id_965b655b-e6e0-442d-8614-e1de6a892684\">\n",
              "    <style>\n",
              "      .colab-df-generate {\n",
              "        background-color: #E8F0FE;\n",
              "        border: none;\n",
              "        border-radius: 50%;\n",
              "        cursor: pointer;\n",
              "        display: none;\n",
              "        fill: #1967D2;\n",
              "        height: 32px;\n",
              "        padding: 0 0 0 0;\n",
              "        width: 32px;\n",
              "      }\n",
              "\n",
              "      .colab-df-generate:hover {\n",
              "        background-color: #E2EBFA;\n",
              "        box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "        fill: #174EA6;\n",
              "      }\n",
              "\n",
              "      [theme=dark] .colab-df-generate {\n",
              "        background-color: #3B4455;\n",
              "        fill: #D2E3FC;\n",
              "      }\n",
              "\n",
              "      [theme=dark] .colab-df-generate:hover {\n",
              "        background-color: #434B5C;\n",
              "        box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "        filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "        fill: #FFFFFF;\n",
              "      }\n",
              "    </style>\n",
              "    <button class=\"colab-df-generate\" onclick=\"generateWithVariable('openai_small_sample_df')\"\n",
              "            title=\"Generate code using this dataframe.\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M7,19H8.4L18.45,9,17,7.55,7,17.6ZM5,21V16.75L18.45,3.32a2,2,0,0,1,2.83,0l1.4,1.43a1.91,1.91,0,0,1,.58,1.4,1.91,1.91,0,0,1-.58,1.4L9.25,21ZM18.45,9,17,7.55Zm-12,3A5.31,5.31,0,0,0,4.9,8.1,5.31,5.31,0,0,0,1,6.5,5.31,5.31,0,0,0,4.9,4.9,5.31,5.31,0,0,0,6.5,1,5.31,5.31,0,0,0,8.1,4.9,5.31,5.31,0,0,0,12,6.5,5.46,5.46,0,0,0,6.5,12Z\"/>\n",
              "  </svg>\n",
              "    </button>\n",
              "    <script>\n",
              "      (() => {\n",
              "      const buttonEl =\n",
              "        document.querySelector('#id_965b655b-e6e0-442d-8614-e1de6a892684 button.colab-df-generate');\n",
              "      buttonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "      buttonEl.onclick = () => {\n",
              "        google.colab.notebook.generateWithVariable('openai_small_sample_df');\n",
              "      }\n",
              "      })();\n",
              "    </script>\n",
              "  </div>\n",
              "\n",
              "    </div>\n",
              "  </div>\n"
            ],
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "dataframe",
              "variable_name": "openai_small_sample_df",
              "summary": "{\n  \"name\": \"openai_small_sample_df\",\n  \"rows\": 2,\n  \"fields\": [\n    {\n      \"column\": \"RecordID\",\n      \"properties\": {\n        \"dtype\": \"number\",\n        \"std\": 127786,\n        \"min\": 274046,\n        \"max\": 454763,\n        \"num_unique_values\": 2,\n        \"samples\": [\n          454763,\n          274046\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    },\n    {\n      \"column\": \"Full job description\",\n      \"properties\": {\n        \"dtype\": \"string\",\n        \"num_unique_values\": 2,\n        \"samples\": [\n          \"FEQ225R94\\n\\nThis role can be remote, targeting candidates on the West Coast or can work in the Pacific Time Zone as needed.\\n\\nAs a Specialist Solutions Architect (SSA) - Data Science & Machine Learning, you will be the trusted technical ML expert to both Databricks customers and the Field Engineering organization. You will work with Solution Architects to guide customers in architecting production-grade ML applications on Databricks, while aligning their technical roadmap with the evolving Databricks Data Intelligence Platform. You will continue to strengthen your technical skills through applying the latest technologies in GenAI, MLOps, and ML, while expanding your impact through mentorship and establishing yourself as a ML expert.\\n\\nThe impact you will have:\\n\\nArchitect production level ML workloads for customers using our unified platform, including end-to-end ML pipelines, training, inference optimization, integration with cloud-native services and MLOps\\nProvide advanced technical support to Solution Architects during the technical sale ranging from feature engineering, training, tracking, serving to model monitoring all within a single platform, and participating in the larger ML SME community in Databricks\\nCollaborate with the product and engineering teams to represent the voice of the customer, define priorities and influence the product roadmap, helping with the adoption of Databricks\\u2019 ML offerings\\nBuild and increase customer data science workloads and apply the best MLOps to productionize these workloads across a variety of domains\\nServe as the trusted technical advisor for customers developing GenAI solutions, such as RAG architectures on enterprise knowledge repos, querying structured data with natural language, content generation, and monitoring\\n\\nWhat we look for:\\n\\n5+ years of hands-on industry ML experience in at least one of the following:\\nML Engineer: Develop production-grade cloud (AWS, Azure, GCP) infrastructure that supports the deployment of ML applications, including drift monitoring\\nData Scientist: Experience with the latest techniques in natural language processing including vector databases, fine-tuning LLMs, and deploying LLMs with tools such as HuggingFace, Langchain, and OpenAI\\nGraduate degree in a quantitative discipline (Computer Science, Engineering, Statistics, Operations Research, etc.) or equivalent practical experience\\nExperience communicating and teaching technical concepts to non-technical and technical audiences alike\\nPassion for collaboration, life-long learning, and driving our values through ML\\n[Preferred] 2+ years customer-facing experience in a pre-sales or post-sales role\\n[Preferred] Experience working with Apache Spark\\u2122 to process large-scale distributed datasets\\nCan meet expectations for technical training and role-specific outcomes within 3 months of hire\\nCan travel up to 30% when needed\\n\\nBenefits\\n\\nMedical, Dental, and Vision\\n401(k) Plan\\nFSA, HSA and Commuter Benefit Plans\\nEquity Awards\\nFlexible Time Off\\nPaid Parental Leave\\nFamily Planning\\nFitness Reimbursement\\nAnnual Career Development Fund\\nHome Office, Work Headphones Reimbursement\\nEmployee Assistance Program (EAP)\\nBusiness Travel Accident Insurance\\nMental Wellness Resources\\n\\nPay Range Transparency\\n\\nDatabricks is committed to fair and equitable compensation practices. The pay range(s) for this role is listed below and represents base salary range for non-commissionable roles or on-target earnings for commissionable roles. Actual compensation packages are based on several factors that are unique to each candidate, including but not limited to job-related skills, depth of experience, relevant certifications and training, and specific work location. Based on the factors above, Databricks utilizes the full width of the range. The total compensation package for this position may also include eligibility for annual performance bonus, equity, and the benefits listed above. For more information regarding which range your location is in visit our page here.\\n\\nZone 1 Pay Range\\n$139,800\\u2014$247,300 USD\\nZone 2 Pay Range\\n$139,800\\u2014$247,300 USD\\nZone 3 Pay Range\\n$139,800\\u2014$247,300 USD\\nZone 4 Pay Range\\n$139,800\\u2014$247,300 USD\\nAbout Databricks\\n\\nDatabricks is the data and AI company. More than 10,000 organizations worldwide \\u2014 including Comcast, Cond\\u00e9 Nast, Grammarly, and over 50% of the Fortune 500 \\u2014 rely on the Databricks Data Intelligence Platform to unify and democratize data, analytics and AI. Databricks is headquartered in San Francisco, with offices around the globe and was founded by the original creators of Lakehouse, Apache Spark\\u2122, Delta Lake and MLflow. To learn more, follow Databricks on Twitter, LinkedIn and Facebook.\\n\\nOur Commitment to Diversity and Inclusion\\n\\nAt Databricks, we are committed to fostering a diverse and inclusive culture where everyone can excel. We take great care to ensure that our hiring practices are inclusive and meet equal employment opportunity standards. Individuals looking for employment at Databricks are considered without regard to age, color, disability, ethnicity, family or marital status, gender identity or expression, language, national origin, physical and mental ability, political affiliation, race, religion, sexual orientation, socio-economic status, veteran status, and other protected characteristics.\\n\\nCompliance\\n\\nIf access to export-controlled technology or source code is required for performance of job duties, it is within Employer's discretion whether to apply for a U.S. government license for such positions, and Employer may decline to proceed with an applicant on this basis alone.\",\n          \"Please return soon to complete your dream job application! Senior Principal Machine Learning Engineer Req id: 39356 Virtual, US OPENTEXT OpenText is a global leader in information management, where innovation, creativity, and collaboration are the key components of our corporate culture. As a member of our team, you will have the opportunity to partner with the most highly regarded companies in the world, tackle complex issues, and contribute to projects that shape the future of digital transformation. The Opportunity: As a Machine Learning Engineer at OpenText, you'll embark on an exhilarating journey, facilitating and creating groundbreaking machine learning models and algorithms and developing software products. Your mission? To unravel intricate business puzzles and craft tailored, one-of-a-kind solutions for our valued clientele. Your canvas? A dynamic lab and sandbox environment where innovation knows no bounds. Collaboration is key as you join forces with diverse teams, weaving machine learning and generative AI magic into the fabric of our enterprise solutions. Your creations will not only meet but exceed the most exacting standards of performance, reliability, and scalability. But this role offers more than just challenges; it's an invitation to help shape the future. By contributing to cutting edge technology, your work will leave an indelible mark on global businesses. Welcome to the forefront of innovation, where the intersection of curiosity, creativity, and technology sparks limitless possibilities. You Are Great At: Collaborate with product managers, software engineers, and data scientists to understand business requirements and translate them into machine learning solutions. Design, develop, and deploy machine learning models and algorithms that enhance the functionality and intelligence of our enterprise software products. Explore and analyze large datasets to identify patterns, trends, and insights that can inform the development of robust machine learning models. Implement and optimize machine learning pipelines for training, testing, and deployment of models in production environments. Work on feature engineering, data preprocessing, and model evaluation to ensure the accuracy and effectiveness of machine learning models. Stay abreast of the latest advancements in machine learning and artificial intelligence, and apply relevant techniques to solve real-world problems. Collaborate with cross-functional teams to integrate machine learning capabilities seamlessly into existing and new software products. Conduct code reviews, provide constructive feedback, and contribute to the continuous improvement of development processes. Maintain documentation for machine learning models, algorithms, and implementation details to ensure transparency and knowledge sharing within the team. Stay informed about industry best practices, emerging trends, and advancements in machine learning, and actively contribute to the company's knowledge base. What It Takes: Required: Hold a Master\\u00e2\\u20ac\\u2122s degree or Ph.D. in Computer Science, Applied Mathematics, Statistics, or a related field with a strong emphasis on Natural Language Processing (NLP) and proficiency in Large Language Models (LLMs). Accumulate 3+ years of professional experience, including a significant time dedicated to relevant and hands-on work in Machine Learning, with a specific focus on NLP and LLMs. Proven track record in developing and deploying advanced Deep Learning models, particularly those tailored for NLP tasks and Large Language Models. Demonstrate substantial knowledge and practical experience in implementing Deep Learning models for extracting insights from textual data, Natural Language Understanding, Text Classification, and Document Analysis, all within the context of NLP and LLMs. Possess excellent communication skills, encompassing both verbal and written proficiency, to effectively convey complex concepts and findings related to NLP and Large Language Models. Proven expertise in utilizing cloud-based environments such as AWS, Azure, or GCP Desired: Solid foundational understanding of machine learning and statistical concepts, with a focus on Natural Language Processing (NLP) and Large Language Models (LLMs). Proficiency in deep learning frameworks commonly used in NLP tasks, such as TensorFlow and PyTorch. Hands-on experience in at least two projects involving Large Language Models and other deep learning applications. A minimum of 5+ years of expertise in Convolutional Neural Networks (CNN) and Recurrent Neural Networks (RNN), particularly in the context of NLP, with the ability to develop custom models for various deep learning tasks. At least 5+ years of practical experience working with popular deep learning software frameworks like TensorFlow and PyTorch. Strong programming skills and a background in building deep learning models using one or more of the following languages: Python, R, Java, C, C++, Scala. Position Type and Expected Hours of Work: Full-time, with occasional evenings or weekends based on client needs. Compensation : This job is expected to pay in the range of $230,000-$250,000 USD in addition to a comprehensive and competitive group benefit and healthcare plan. Individual compensation will be determined based on skills and experience comparable to the job requirements. OpenText's commitment to diversity and inclusion surpasses legal requirements, evident in our Equal Employment Opportunity Statement of Policy which promotes a respectful and empowering environment for employees of all backgrounds, culture, national origin, race, color, gender, gender identification, sexual orientation, family status, age, veteran status, disability, religion, or other basis protected by applicable laws. If you need assistance and, or a reasonable accommodation due to a disability during the application or recruiting process, please contact us at hr@opentext.com. Our proactive approach fosters collaboration, innovation, and personal growth, enriching OpenText's vibrant workplace.\"\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    }\n  ]\n}"
            }
          },
          "metadata": {},
          "execution_count": 12
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Set up prompts\n",
        "---"
      ],
      "metadata": {
        "id": "qTIOKHloslZk"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Extraction prompt\n",
        "ext_prompt = f\"\"\"Extract all AI product and project names from the following job\n",
        "description.\n",
        "\n",
        "Examples of what to look for include Kubernetes, Docker,\n",
        "MongoDB, PostgreSQL, BigQuery, Llama, GPT, Phi, Spark, Dataflow, JAX,\n",
        "Bedrock, Vertex AI,  etc.\n",
        "\n",
        "Only return these examples if they appear in the description below.\n",
        "Also, please be thorough and return as many products and/or technologies\n",
        "as possible. Lastly, return a comma-delimited list, nothing more:\n",
        "\n",
        "Description: \\n\"\"\"\n"
      ],
      "metadata": {
        "id": "er8fmBI5oqT1"
      },
      "execution_count": 13,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# Categorization prompt\n",
        "cat_prompt = f\"\"\"\n",
        "Categorize each of the following technologies into one of the\n",
        "following categories. Only return one Category; do not leave any blank.\n",
        "\n",
        "Categories:\n",
        "1. Infrastructure Layer\n",
        "2. Data Layer\n",
        "3. Model Training Layer\n",
        "4. Model Serving Layer\n",
        "5. Pipeline and Automation Layer\n",
        "6. Monitoring and Management Layer\n",
        "7. User Interface Layer\n",
        "8. Security and Compliance Layer\n",
        "9. Collaboration and Communications Layer\n",
        "\n",
        "Output format:\n",
        "Technology: Category\n",
        "\n",
        "Technologies:\n",
        "\\n\"\"\""
      ],
      "metadata": {
        "id": "-WEIdoasswP_"
      },
      "execution_count": 14,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "---\n",
        "## Test with OpenAI"
      ],
      "metadata": {
        "id": "mucxNA6DqabX"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# define function to extract technologies from job desc\n",
        "def extract_technologies(description):\n",
        "    \"\"\"Generate list of technologies extracted from text\"\"\"\n",
        "    prompt = ext_prompt + description\n",
        "    response = openai_client.completions.create(\n",
        "        model=openai_model,\n",
        "        prompt=prompt,\n",
        "    )\n",
        "    return response.choices[0].text.strip()"
      ],
      "metadata": {
        "id": "CZ2Xt5Bn3zKl"
      },
      "execution_count": 15,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# Run sample through OpenAI for extraction\n",
        "openai_small_sample_df['Extracted Technologies'] = openai_small_sample_df['Full job description'].apply(extract_technologies)\n"
      ],
      "metadata": {
        "id": "P_86uh7799If"
      },
      "execution_count": 16,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "openai_small_sample_df"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1525
        },
        "id": "xwq-nVdv5IRE",
        "outputId": "5e69fdfc-1504-462f-804b-59b34aa6f79d"
      },
      "execution_count": 17,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "        RecordID                               Full job description  \\\n",
              "274043    274046  Please return soon to complete your dream job ...   \n",
              "454763    454763  FEQ225R94\\n\\nThis role can be remote, targetin...   \n",
              "\n",
              "                                   Extracted Technologies  \n",
              "274043  Kubernetes, OpenText, information management, ...  \n",
              "454763                  Databricks, GenAI, MLOps, ML, RAG  "
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-5b23ec72-2161-4f82-ab21-dff80f7ce65d\" class=\"colab-df-container\">\n",
              "    <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>RecordID</th>\n",
              "      <th>Full job description</th>\n",
              "      <th>Extracted Technologies</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>274043</th>\n",
              "      <td>274046</td>\n",
              "      <td>Please return soon to complete your dream job ...</td>\n",
              "      <td>Kubernetes, OpenText, information management, ...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>454763</th>\n",
              "      <td>454763</td>\n",
              "      <td>FEQ225R94\\n\\nThis role can be remote, targetin...</td>\n",
              "      <td>Databricks, GenAI, MLOps, ML, RAG</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "    <div class=\"colab-df-buttons\">\n",
              "\n",
              "  <div class=\"colab-df-container\">\n",
              "    <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-5b23ec72-2161-4f82-ab21-dff80f7ce65d')\"\n",
              "            title=\"Convert this dataframe to an interactive table.\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\" viewBox=\"0 -960 960 960\">\n",
              "    <path d=\"M120-120v-720h720v720H120Zm60-500h600v-160H180v160Zm220 220h160v-160H400v160Zm0 220h160v-160H400v160ZM180-400h160v-160H180v160Zm440 0h160v-160H620v160ZM180-180h160v-160H180v160Zm440 0h160v-160H620v160Z\"/>\n",
              "  </svg>\n",
              "    </button>\n",
              "\n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    .colab-df-buttons div {\n",
              "      margin-bottom: 4px;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "    <script>\n",
              "      const buttonEl =\n",
              "        document.querySelector('#df-5b23ec72-2161-4f82-ab21-dff80f7ce65d button.colab-df-convert');\n",
              "      buttonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "      async function convertToInteractive(key) {\n",
              "        const element = document.querySelector('#df-5b23ec72-2161-4f82-ab21-dff80f7ce65d');\n",
              "        const dataTable =\n",
              "          await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                    [key], {});\n",
              "        if (!dataTable) return;\n",
              "\n",
              "        const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "          '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "          + ' to learn more about interactive tables.';\n",
              "        element.innerHTML = '';\n",
              "        dataTable['output_type'] = 'display_data';\n",
              "        await google.colab.output.renderOutput(dataTable, element);\n",
              "        const docLink = document.createElement('div');\n",
              "        docLink.innerHTML = docLinkHtml;\n",
              "        element.appendChild(docLink);\n",
              "      }\n",
              "    </script>\n",
              "  </div>\n",
              "\n",
              "\n",
              "<div id=\"df-10db24ca-eed4-431b-9f7c-4cb00d44c96f\">\n",
              "  <button class=\"colab-df-quickchart\" onclick=\"quickchart('df-10db24ca-eed4-431b-9f7c-4cb00d44c96f')\"\n",
              "            title=\"Suggest charts\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "<svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "     width=\"24px\">\n",
              "    <g>\n",
              "        <path d=\"M19 3H5c-1.1 0-2 .9-2 2v14c0 1.1.9 2 2 2h14c1.1 0 2-.9 2-2V5c0-1.1-.9-2-2-2zM9 17H7v-7h2v7zm4 0h-2V7h2v10zm4 0h-2v-4h2v4z\"/>\n",
              "    </g>\n",
              "</svg>\n",
              "  </button>\n",
              "\n",
              "<style>\n",
              "  .colab-df-quickchart {\n",
              "      --bg-color: #E8F0FE;\n",
              "      --fill-color: #1967D2;\n",
              "      --hover-bg-color: #E2EBFA;\n",
              "      --hover-fill-color: #174EA6;\n",
              "      --disabled-fill-color: #AAA;\n",
              "      --disabled-bg-color: #DDD;\n",
              "  }\n",
              "\n",
              "  [theme=dark] .colab-df-quickchart {\n",
              "      --bg-color: #3B4455;\n",
              "      --fill-color: #D2E3FC;\n",
              "      --hover-bg-color: #434B5C;\n",
              "      --hover-fill-color: #FFFFFF;\n",
              "      --disabled-bg-color: #3B4455;\n",
              "      --disabled-fill-color: #666;\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart {\n",
              "    background-color: var(--bg-color);\n",
              "    border: none;\n",
              "    border-radius: 50%;\n",
              "    cursor: pointer;\n",
              "    display: none;\n",
              "    fill: var(--fill-color);\n",
              "    height: 32px;\n",
              "    padding: 0;\n",
              "    width: 32px;\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart:hover {\n",
              "    background-color: var(--hover-bg-color);\n",
              "    box-shadow: 0 1px 2px rgba(60, 64, 67, 0.3), 0 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "    fill: var(--button-hover-fill-color);\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart-complete:disabled,\n",
              "  .colab-df-quickchart-complete:disabled:hover {\n",
              "    background-color: var(--disabled-bg-color);\n",
              "    fill: var(--disabled-fill-color);\n",
              "    box-shadow: none;\n",
              "  }\n",
              "\n",
              "  .colab-df-spinner {\n",
              "    border: 2px solid var(--fill-color);\n",
              "    border-color: transparent;\n",
              "    border-bottom-color: var(--fill-color);\n",
              "    animation:\n",
              "      spin 1s steps(1) infinite;\n",
              "  }\n",
              "\n",
              "  @keyframes spin {\n",
              "    0% {\n",
              "      border-color: transparent;\n",
              "      border-bottom-color: var(--fill-color);\n",
              "      border-left-color: var(--fill-color);\n",
              "    }\n",
              "    20% {\n",
              "      border-color: transparent;\n",
              "      border-left-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "    }\n",
              "    30% {\n",
              "      border-color: transparent;\n",
              "      border-left-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "      border-right-color: var(--fill-color);\n",
              "    }\n",
              "    40% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "    }\n",
              "    60% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "    }\n",
              "    80% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "      border-bottom-color: var(--fill-color);\n",
              "    }\n",
              "    90% {\n",
              "      border-color: transparent;\n",
              "      border-bottom-color: var(--fill-color);\n",
              "    }\n",
              "  }\n",
              "</style>\n",
              "\n",
              "  <script>\n",
              "    async function quickchart(key) {\n",
              "      const quickchartButtonEl =\n",
              "        document.querySelector('#' + key + ' button');\n",
              "      quickchartButtonEl.disabled = true;  // To prevent multiple clicks.\n",
              "      quickchartButtonEl.classList.add('colab-df-spinner');\n",
              "      try {\n",
              "        const charts = await google.colab.kernel.invokeFunction(\n",
              "            'suggestCharts', [key], {});\n",
              "      } catch (error) {\n",
              "        console.error('Error during call to suggestCharts:', error);\n",
              "      }\n",
              "      quickchartButtonEl.classList.remove('colab-df-spinner');\n",
              "      quickchartButtonEl.classList.add('colab-df-quickchart-complete');\n",
              "    }\n",
              "    (() => {\n",
              "      let quickchartButtonEl =\n",
              "        document.querySelector('#df-10db24ca-eed4-431b-9f7c-4cb00d44c96f button');\n",
              "      quickchartButtonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "    })();\n",
              "  </script>\n",
              "</div>\n",
              "\n",
              "  <div id=\"id_4c5ec2c9-7684-456c-bdd4-68c37abbc28e\">\n",
              "    <style>\n",
              "      .colab-df-generate {\n",
              "        background-color: #E8F0FE;\n",
              "        border: none;\n",
              "        border-radius: 50%;\n",
              "        cursor: pointer;\n",
              "        display: none;\n",
              "        fill: #1967D2;\n",
              "        height: 32px;\n",
              "        padding: 0 0 0 0;\n",
              "        width: 32px;\n",
              "      }\n",
              "\n",
              "      .colab-df-generate:hover {\n",
              "        background-color: #E2EBFA;\n",
              "        box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "        fill: #174EA6;\n",
              "      }\n",
              "\n",
              "      [theme=dark] .colab-df-generate {\n",
              "        background-color: #3B4455;\n",
              "        fill: #D2E3FC;\n",
              "      }\n",
              "\n",
              "      [theme=dark] .colab-df-generate:hover {\n",
              "        background-color: #434B5C;\n",
              "        box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "        filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "        fill: #FFFFFF;\n",
              "      }\n",
              "    </style>\n",
              "    <button class=\"colab-df-generate\" onclick=\"generateWithVariable('openai_small_sample_df')\"\n",
              "            title=\"Generate code using this dataframe.\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M7,19H8.4L18.45,9,17,7.55,7,17.6ZM5,21V16.75L18.45,3.32a2,2,0,0,1,2.83,0l1.4,1.43a1.91,1.91,0,0,1,.58,1.4,1.91,1.91,0,0,1-.58,1.4L9.25,21ZM18.45,9,17,7.55Zm-12,3A5.31,5.31,0,0,0,4.9,8.1,5.31,5.31,0,0,0,1,6.5,5.31,5.31,0,0,0,4.9,4.9,5.31,5.31,0,0,0,6.5,1,5.31,5.31,0,0,0,8.1,4.9,5.31,5.31,0,0,0,12,6.5,5.46,5.46,0,0,0,6.5,12Z\"/>\n",
              "  </svg>\n",
              "    </button>\n",
              "    <script>\n",
              "      (() => {\n",
              "      const buttonEl =\n",
              "        document.querySelector('#id_4c5ec2c9-7684-456c-bdd4-68c37abbc28e button.colab-df-generate');\n",
              "      buttonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "      buttonEl.onclick = () => {\n",
              "        google.colab.notebook.generateWithVariable('openai_small_sample_df');\n",
              "      }\n",
              "      })();\n",
              "    </script>\n",
              "  </div>\n",
              "\n",
              "    </div>\n",
              "  </div>\n"
            ],
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "dataframe",
              "variable_name": "openai_small_sample_df",
              "summary": "{\n  \"name\": \"openai_small_sample_df\",\n  \"rows\": 2,\n  \"fields\": [\n    {\n      \"column\": \"RecordID\",\n      \"properties\": {\n        \"dtype\": \"number\",\n        \"std\": 127786,\n        \"min\": 274046,\n        \"max\": 454763,\n        \"num_unique_values\": 2,\n        \"samples\": [\n          454763,\n          274046\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    },\n    {\n      \"column\": \"Full job description\",\n      \"properties\": {\n        \"dtype\": \"string\",\n        \"num_unique_values\": 2,\n        \"samples\": [\n          \"FEQ225R94\\n\\nThis role can be remote, targeting candidates on the West Coast or can work in the Pacific Time Zone as needed.\\n\\nAs a Specialist Solutions Architect (SSA) - Data Science & Machine Learning, you will be the trusted technical ML expert to both Databricks customers and the Field Engineering organization. You will work with Solution Architects to guide customers in architecting production-grade ML applications on Databricks, while aligning their technical roadmap with the evolving Databricks Data Intelligence Platform. You will continue to strengthen your technical skills through applying the latest technologies in GenAI, MLOps, and ML, while expanding your impact through mentorship and establishing yourself as a ML expert.\\n\\nThe impact you will have:\\n\\nArchitect production level ML workloads for customers using our unified platform, including end-to-end ML pipelines, training, inference optimization, integration with cloud-native services and MLOps\\nProvide advanced technical support to Solution Architects during the technical sale ranging from feature engineering, training, tracking, serving to model monitoring all within a single platform, and participating in the larger ML SME community in Databricks\\nCollaborate with the product and engineering teams to represent the voice of the customer, define priorities and influence the product roadmap, helping with the adoption of Databricks\\u2019 ML offerings\\nBuild and increase customer data science workloads and apply the best MLOps to productionize these workloads across a variety of domains\\nServe as the trusted technical advisor for customers developing GenAI solutions, such as RAG architectures on enterprise knowledge repos, querying structured data with natural language, content generation, and monitoring\\n\\nWhat we look for:\\n\\n5+ years of hands-on industry ML experience in at least one of the following:\\nML Engineer: Develop production-grade cloud (AWS, Azure, GCP) infrastructure that supports the deployment of ML applications, including drift monitoring\\nData Scientist: Experience with the latest techniques in natural language processing including vector databases, fine-tuning LLMs, and deploying LLMs with tools such as HuggingFace, Langchain, and OpenAI\\nGraduate degree in a quantitative discipline (Computer Science, Engineering, Statistics, Operations Research, etc.) or equivalent practical experience\\nExperience communicating and teaching technical concepts to non-technical and technical audiences alike\\nPassion for collaboration, life-long learning, and driving our values through ML\\n[Preferred] 2+ years customer-facing experience in a pre-sales or post-sales role\\n[Preferred] Experience working with Apache Spark\\u2122 to process large-scale distributed datasets\\nCan meet expectations for technical training and role-specific outcomes within 3 months of hire\\nCan travel up to 30% when needed\\n\\nBenefits\\n\\nMedical, Dental, and Vision\\n401(k) Plan\\nFSA, HSA and Commuter Benefit Plans\\nEquity Awards\\nFlexible Time Off\\nPaid Parental Leave\\nFamily Planning\\nFitness Reimbursement\\nAnnual Career Development Fund\\nHome Office, Work Headphones Reimbursement\\nEmployee Assistance Program (EAP)\\nBusiness Travel Accident Insurance\\nMental Wellness Resources\\n\\nPay Range Transparency\\n\\nDatabricks is committed to fair and equitable compensation practices. The pay range(s) for this role is listed below and represents base salary range for non-commissionable roles or on-target earnings for commissionable roles. Actual compensation packages are based on several factors that are unique to each candidate, including but not limited to job-related skills, depth of experience, relevant certifications and training, and specific work location. Based on the factors above, Databricks utilizes the full width of the range. The total compensation package for this position may also include eligibility for annual performance bonus, equity, and the benefits listed above. For more information regarding which range your location is in visit our page here.\\n\\nZone 1 Pay Range\\n$139,800\\u2014$247,300 USD\\nZone 2 Pay Range\\n$139,800\\u2014$247,300 USD\\nZone 3 Pay Range\\n$139,800\\u2014$247,300 USD\\nZone 4 Pay Range\\n$139,800\\u2014$247,300 USD\\nAbout Databricks\\n\\nDatabricks is the data and AI company. More than 10,000 organizations worldwide \\u2014 including Comcast, Cond\\u00e9 Nast, Grammarly, and over 50% of the Fortune 500 \\u2014 rely on the Databricks Data Intelligence Platform to unify and democratize data, analytics and AI. Databricks is headquartered in San Francisco, with offices around the globe and was founded by the original creators of Lakehouse, Apache Spark\\u2122, Delta Lake and MLflow. To learn more, follow Databricks on Twitter, LinkedIn and Facebook.\\n\\nOur Commitment to Diversity and Inclusion\\n\\nAt Databricks, we are committed to fostering a diverse and inclusive culture where everyone can excel. We take great care to ensure that our hiring practices are inclusive and meet equal employment opportunity standards. Individuals looking for employment at Databricks are considered without regard to age, color, disability, ethnicity, family or marital status, gender identity or expression, language, national origin, physical and mental ability, political affiliation, race, religion, sexual orientation, socio-economic status, veteran status, and other protected characteristics.\\n\\nCompliance\\n\\nIf access to export-controlled technology or source code is required for performance of job duties, it is within Employer's discretion whether to apply for a U.S. government license for such positions, and Employer may decline to proceed with an applicant on this basis alone.\",\n          \"Please return soon to complete your dream job application! Senior Principal Machine Learning Engineer Req id: 39356 Virtual, US OPENTEXT OpenText is a global leader in information management, where innovation, creativity, and collaboration are the key components of our corporate culture. As a member of our team, you will have the opportunity to partner with the most highly regarded companies in the world, tackle complex issues, and contribute to projects that shape the future of digital transformation. The Opportunity: As a Machine Learning Engineer at OpenText, you'll embark on an exhilarating journey, facilitating and creating groundbreaking machine learning models and algorithms and developing software products. Your mission? To unravel intricate business puzzles and craft tailored, one-of-a-kind solutions for our valued clientele. Your canvas? A dynamic lab and sandbox environment where innovation knows no bounds. Collaboration is key as you join forces with diverse teams, weaving machine learning and generative AI magic into the fabric of our enterprise solutions. Your creations will not only meet but exceed the most exacting standards of performance, reliability, and scalability. But this role offers more than just challenges; it's an invitation to help shape the future. By contributing to cutting edge technology, your work will leave an indelible mark on global businesses. Welcome to the forefront of innovation, where the intersection of curiosity, creativity, and technology sparks limitless possibilities. You Are Great At: Collaborate with product managers, software engineers, and data scientists to understand business requirements and translate them into machine learning solutions. Design, develop, and deploy machine learning models and algorithms that enhance the functionality and intelligence of our enterprise software products. Explore and analyze large datasets to identify patterns, trends, and insights that can inform the development of robust machine learning models. Implement and optimize machine learning pipelines for training, testing, and deployment of models in production environments. Work on feature engineering, data preprocessing, and model evaluation to ensure the accuracy and effectiveness of machine learning models. Stay abreast of the latest advancements in machine learning and artificial intelligence, and apply relevant techniques to solve real-world problems. Collaborate with cross-functional teams to integrate machine learning capabilities seamlessly into existing and new software products. Conduct code reviews, provide constructive feedback, and contribute to the continuous improvement of development processes. Maintain documentation for machine learning models, algorithms, and implementation details to ensure transparency and knowledge sharing within the team. Stay informed about industry best practices, emerging trends, and advancements in machine learning, and actively contribute to the company's knowledge base. What It Takes: Required: Hold a Master\\u00e2\\u20ac\\u2122s degree or Ph.D. in Computer Science, Applied Mathematics, Statistics, or a related field with a strong emphasis on Natural Language Processing (NLP) and proficiency in Large Language Models (LLMs). Accumulate 3+ years of professional experience, including a significant time dedicated to relevant and hands-on work in Machine Learning, with a specific focus on NLP and LLMs. Proven track record in developing and deploying advanced Deep Learning models, particularly those tailored for NLP tasks and Large Language Models. Demonstrate substantial knowledge and practical experience in implementing Deep Learning models for extracting insights from textual data, Natural Language Understanding, Text Classification, and Document Analysis, all within the context of NLP and LLMs. Possess excellent communication skills, encompassing both verbal and written proficiency, to effectively convey complex concepts and findings related to NLP and Large Language Models. Proven expertise in utilizing cloud-based environments such as AWS, Azure, or GCP Desired: Solid foundational understanding of machine learning and statistical concepts, with a focus on Natural Language Processing (NLP) and Large Language Models (LLMs). Proficiency in deep learning frameworks commonly used in NLP tasks, such as TensorFlow and PyTorch. Hands-on experience in at least two projects involving Large Language Models and other deep learning applications. A minimum of 5+ years of expertise in Convolutional Neural Networks (CNN) and Recurrent Neural Networks (RNN), particularly in the context of NLP, with the ability to develop custom models for various deep learning tasks. At least 5+ years of practical experience working with popular deep learning software frameworks like TensorFlow and PyTorch. Strong programming skills and a background in building deep learning models using one or more of the following languages: Python, R, Java, C, C++, Scala. Position Type and Expected Hours of Work: Full-time, with occasional evenings or weekends based on client needs. Compensation : This job is expected to pay in the range of $230,000-$250,000 USD in addition to a comprehensive and competitive group benefit and healthcare plan. Individual compensation will be determined based on skills and experience comparable to the job requirements. OpenText's commitment to diversity and inclusion surpasses legal requirements, evident in our Equal Employment Opportunity Statement of Policy which promotes a respectful and empowering environment for employees of all backgrounds, culture, national origin, race, color, gender, gender identification, sexual orientation, family status, age, veteran status, disability, religion, or other basis protected by applicable laws. If you need assistance and, or a reasonable accommodation due to a disability during the application or recruiting process, please contact us at hr@opentext.com. Our proactive approach fosters collaboration, innovation, and personal growth, enriching OpenText's vibrant workplace.\"\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    },\n    {\n      \"column\": \"Extracted Technologies\",\n      \"properties\": {\n        \"dtype\": \"string\",\n        \"num_unique_values\": 2,\n        \"samples\": [\n          \"Databricks, GenAI, MLOps, ML, RAG\",\n          \"Kubernetes, OpenText, information management, innovation, collaboration, virtual,\"\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    }\n  ]\n}"
            }
          },
          "metadata": {},
          "execution_count": 17
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Now let's categorize these\n",
        "def categorize_technologies(technologies):\n",
        "    \"\"\"categorize each of the technologies into one of the\n",
        "    following categories\"\"\"\n",
        "    prompt = cat_prompt + technologies\n",
        "    response = openai_client.completions.create(\n",
        "        model=openai_model,\n",
        "        prompt=prompt,\n",
        "    )\n",
        "    return response.choices[0].text.strip()"
      ],
      "metadata": {
        "id": "MkoVTdLN9V24"
      },
      "execution_count": 18,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# Now we parse categorized output\n",
        "def parse_categorized_output(categorized_output):\n",
        "    \"\"\"parse categorized output\"\"\"\n",
        "    tech_list = categorized_output.split('\\n')\n",
        "    parsed_tech = []\n",
        "\n",
        "    for tech in tech_list:\n",
        "        if ': ' in tech:\n",
        "            technology, category = tech.split(': ')\n",
        "            parsed_tech.append({'Technology': technology, 'Layer': category})\n",
        "\n",
        "    return parsed_tech\n"
      ],
      "metadata": {
        "id": "jmjcYseb9V25"
      },
      "execution_count": 19,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# Now we categorize our technologies\n",
        "openai_small_sample_df['Categorized Technologies'] = openai_small_sample_df['Extracted Technologies'].apply(categorize_technologies)\n"
      ],
      "metadata": {
        "id": "lBY6T2Zu9V25"
      },
      "execution_count": 20,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# Finally, we parse all of this and put it in a new df\n",
        "categorized_tech_list = []\n",
        "\n",
        "for _, row in openai_small_sample_df.iterrows():\n",
        "    categorized_output = row['Categorized Technologies']\n",
        "    parsed_output = parse_categorized_output(categorized_output)\n",
        "    for entry in parsed_output:\n",
        "        entry['RecordID'] = row['RecordID']\n",
        "        categorized_tech_list.append(entry)\n",
        "\n",
        "categorized_tech_df = pd.DataFrame(categorized_tech_list)\n",
        "\n",
        "# Integrate the categorized technologies back with the original DataFrame\n",
        "final_df = df.merge(categorized_tech_df, on='RecordID', how='left')\n"
      ],
      "metadata": {
        "id": "udwLrVOS9V26"
      },
      "execution_count": 21,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "final_df.columns"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "a9c08083-86e7-4476-c069-39861b27ad3d",
        "id": "NZlwiy2o9V26"
      },
      "execution_count": 22,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Index(['RecordID', 'Job title', 'Full job description', 'Date collected',\n",
              "       'Date posted', 'Week', 'Year', 'Month', 'Quarter', 'Hiring entity',\n",
              "       'Entity size', 'Home office hybrid work', 'Full or part time status',\n",
              "       'Number of applicants', 'Longitude', 'Latitude', 'City', 'Territory',\n",
              "       'Region', 'Experience level', 'Education level', 'Salary start',\n",
              "       'Salary end', 'Main industry', 'Job title category', 'AI technologies',\n",
              "       'AI skills', 'Source name', 'Source URL', 'File Date', 'Technology',\n",
              "       'Layer'],\n",
              "      dtype='object')"
            ]
          },
          "metadata": {},
          "execution_count": 22
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "final_df.shape"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "7be74fcd-8645-4ffd-f31f-c93ab159a3a5",
        "id": "wmqTpQEB9V27"
      },
      "execution_count": 23,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(205517, 32)"
            ]
          },
          "metadata": {},
          "execution_count": 23
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# prompt: find and remove rows where Technology and Layer are both NaN values\n",
        "final_df_deduped = final_df.dropna(subset=['Technology', 'Layer'])\n"
      ],
      "metadata": {
        "id": "teoFY-ls9V27"
      },
      "execution_count": 24,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "final_df_deduped.shape"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "9309c484-a76a-48b6-c1cc-dcc4c3c8ddc7",
        "id": "QcJ_bP529V27"
      },
      "execution_count": 25,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(2, 32)"
            ]
          },
          "metadata": {},
          "execution_count": 25
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "final_df['Layer'].value_counts()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "1dcfc066-2432-4ac1-852d-d3d420d02620",
        "id": "rkmJu1bs9V28"
      },
      "execution_count": 26,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Layer\n",
              "Infrastructure Layer    2\n",
              "Name: count, dtype: int64"
            ]
          },
          "metadata": {},
          "execution_count": 26
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "final_df['Technology'].value_counts()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "746492f7-803f-4fb1-b3dd-730a5b1807d4",
        "id": "6am9-Tgy9V29"
      },
      "execution_count": 27,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Technology\n",
              "Kubernetes    1\n",
              "Databricks    1\n",
              "Name: count, dtype: int64"
            ]
          },
          "metadata": {},
          "execution_count": 27
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# prompt: Create a chart that shows the correlation between Layer and Technology\n",
        "correlation_matrix = final_df_deduped.groupby(['Layer', 'Technology']).size().unstack()\n",
        "\n",
        "# Create a heatmap of the correlation matrix\n",
        "plt.figure(figsize=(5, 5))\n",
        "sns.heatmap(correlation_matrix, annot=True, cmap='coolwarm')\n",
        "plt.xlabel('Technology')\n",
        "plt.ylabel('Layer')\n",
        "plt.title('Correlation between Layer and Technology')\n",
        "plt.show()\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 487
        },
        "outputId": "032d33f1-6a35-4662-d265-a5df97f67d61",
        "id": "tdtOsS4I9V29"
      },
      "execution_count": 28,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 500x500 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Same experiment this time with Anthropic Claude"
      ],
      "metadata": {
        "id": "kXiZtGPCnP1Y"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Function to extract product and technology names\n",
        "def extract_technologies(description):\n",
        "    \"\"\"generate list of technologies extracted\n",
        "    from text\"\"\"\n",
        "    message = anthropic_client.messages.create(\n",
        "        model=anthropic_model,\n",
        "        max_tokens=1024,\n",
        "        messages=[\n",
        "            {\n",
        "                \"role\": \"user\",\n",
        "                \"content\": ext_prompt + description\n",
        "            },\n",
        "        ]\n",
        "    ).content[0].text\n",
        "    return message\n"
      ],
      "metadata": {
        "id": "u0l-XkIBPJ_B"
      },
      "execution_count": 29,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "anthropic_small_sample_df['Extracted Technologies'] = anthropic_small_sample_df['Full job description'].apply(extract_technologies)\n"
      ],
      "metadata": {
        "id": "OIrKQ8xaq6VZ"
      },
      "execution_count": 30,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "anthropic_small_sample_df"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1294
        },
        "id": "IVkieqjm7WhX",
        "outputId": "79628e2e-1aa4-4bef-e934-0cd624a9fdd5"
      },
      "execution_count": 31,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "        RecordID                               Full job description  \\\n",
              "274043    274046  Please return soon to complete your dream job ...   \n",
              "454763    454763  FEQ225R94\\n\\nThis role can be remote, targetin...   \n",
              "\n",
              "                                   Extracted Technologies  \n",
              "274043  Kubernetes, Docker, MongoDB, PostgreSQL, BigQu...  \n",
              "454763  Apache Spark, Delta Lake, MLflow, Databricks, ...  "
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-80d9b77d-3a84-420e-ae72-0e993fa77aa4\" class=\"colab-df-container\">\n",
              "    <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>RecordID</th>\n",
              "      <th>Full job description</th>\n",
              "      <th>Extracted Technologies</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>274043</th>\n",
              "      <td>274046</td>\n",
              "      <td>Please return soon to complete your dream job ...</td>\n",
              "      <td>Kubernetes, Docker, MongoDB, PostgreSQL, BigQu...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>454763</th>\n",
              "      <td>454763</td>\n",
              "      <td>FEQ225R94\\n\\nThis role can be remote, targetin...</td>\n",
              "      <td>Apache Spark, Delta Lake, MLflow, Databricks, ...</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "    <div class=\"colab-df-buttons\">\n",
              "\n",
              "  <div class=\"colab-df-container\">\n",
              "    <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-80d9b77d-3a84-420e-ae72-0e993fa77aa4')\"\n",
              "            title=\"Convert this dataframe to an interactive table.\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\" viewBox=\"0 -960 960 960\">\n",
              "    <path d=\"M120-120v-720h720v720H120Zm60-500h600v-160H180v160Zm220 220h160v-160H400v160Zm0 220h160v-160H400v160ZM180-400h160v-160H180v160Zm440 0h160v-160H620v160ZM180-180h160v-160H180v160Zm440 0h160v-160H620v160Z\"/>\n",
              "  </svg>\n",
              "    </button>\n",
              "\n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    .colab-df-buttons div {\n",
              "      margin-bottom: 4px;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "    <script>\n",
              "      const buttonEl =\n",
              "        document.querySelector('#df-80d9b77d-3a84-420e-ae72-0e993fa77aa4 button.colab-df-convert');\n",
              "      buttonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "      async function convertToInteractive(key) {\n",
              "        const element = document.querySelector('#df-80d9b77d-3a84-420e-ae72-0e993fa77aa4');\n",
              "        const dataTable =\n",
              "          await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                    [key], {});\n",
              "        if (!dataTable) return;\n",
              "\n",
              "        const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "          '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "          + ' to learn more about interactive tables.';\n",
              "        element.innerHTML = '';\n",
              "        dataTable['output_type'] = 'display_data';\n",
              "        await google.colab.output.renderOutput(dataTable, element);\n",
              "        const docLink = document.createElement('div');\n",
              "        docLink.innerHTML = docLinkHtml;\n",
              "        element.appendChild(docLink);\n",
              "      }\n",
              "    </script>\n",
              "  </div>\n",
              "\n",
              "\n",
              "<div id=\"df-3317b508-e0f2-4b8c-8224-19bdfcedc09f\">\n",
              "  <button class=\"colab-df-quickchart\" onclick=\"quickchart('df-3317b508-e0f2-4b8c-8224-19bdfcedc09f')\"\n",
              "            title=\"Suggest charts\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "<svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "     width=\"24px\">\n",
              "    <g>\n",
              "        <path d=\"M19 3H5c-1.1 0-2 .9-2 2v14c0 1.1.9 2 2 2h14c1.1 0 2-.9 2-2V5c0-1.1-.9-2-2-2zM9 17H7v-7h2v7zm4 0h-2V7h2v10zm4 0h-2v-4h2v4z\"/>\n",
              "    </g>\n",
              "</svg>\n",
              "  </button>\n",
              "\n",
              "<style>\n",
              "  .colab-df-quickchart {\n",
              "      --bg-color: #E8F0FE;\n",
              "      --fill-color: #1967D2;\n",
              "      --hover-bg-color: #E2EBFA;\n",
              "      --hover-fill-color: #174EA6;\n",
              "      --disabled-fill-color: #AAA;\n",
              "      --disabled-bg-color: #DDD;\n",
              "  }\n",
              "\n",
              "  [theme=dark] .colab-df-quickchart {\n",
              "      --bg-color: #3B4455;\n",
              "      --fill-color: #D2E3FC;\n",
              "      --hover-bg-color: #434B5C;\n",
              "      --hover-fill-color: #FFFFFF;\n",
              "      --disabled-bg-color: #3B4455;\n",
              "      --disabled-fill-color: #666;\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart {\n",
              "    background-color: var(--bg-color);\n",
              "    border: none;\n",
              "    border-radius: 50%;\n",
              "    cursor: pointer;\n",
              "    display: none;\n",
              "    fill: var(--fill-color);\n",
              "    height: 32px;\n",
              "    padding: 0;\n",
              "    width: 32px;\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart:hover {\n",
              "    background-color: var(--hover-bg-color);\n",
              "    box-shadow: 0 1px 2px rgba(60, 64, 67, 0.3), 0 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "    fill: var(--button-hover-fill-color);\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart-complete:disabled,\n",
              "  .colab-df-quickchart-complete:disabled:hover {\n",
              "    background-color: var(--disabled-bg-color);\n",
              "    fill: var(--disabled-fill-color);\n",
              "    box-shadow: none;\n",
              "  }\n",
              "\n",
              "  .colab-df-spinner {\n",
              "    border: 2px solid var(--fill-color);\n",
              "    border-color: transparent;\n",
              "    border-bottom-color: var(--fill-color);\n",
              "    animation:\n",
              "      spin 1s steps(1) infinite;\n",
              "  }\n",
              "\n",
              "  @keyframes spin {\n",
              "    0% {\n",
              "      border-color: transparent;\n",
              "      border-bottom-color: var(--fill-color);\n",
              "      border-left-color: var(--fill-color);\n",
              "    }\n",
              "    20% {\n",
              "      border-color: transparent;\n",
              "      border-left-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "    }\n",
              "    30% {\n",
              "      border-color: transparent;\n",
              "      border-left-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "      border-right-color: var(--fill-color);\n",
              "    }\n",
              "    40% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "    }\n",
              "    60% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "    }\n",
              "    80% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "      border-bottom-color: var(--fill-color);\n",
              "    }\n",
              "    90% {\n",
              "      border-color: transparent;\n",
              "      border-bottom-color: var(--fill-color);\n",
              "    }\n",
              "  }\n",
              "</style>\n",
              "\n",
              "  <script>\n",
              "    async function quickchart(key) {\n",
              "      const quickchartButtonEl =\n",
              "        document.querySelector('#' + key + ' button');\n",
              "      quickchartButtonEl.disabled = true;  // To prevent multiple clicks.\n",
              "      quickchartButtonEl.classList.add('colab-df-spinner');\n",
              "      try {\n",
              "        const charts = await google.colab.kernel.invokeFunction(\n",
              "            'suggestCharts', [key], {});\n",
              "      } catch (error) {\n",
              "        console.error('Error during call to suggestCharts:', error);\n",
              "      }\n",
              "      quickchartButtonEl.classList.remove('colab-df-spinner');\n",
              "      quickchartButtonEl.classList.add('colab-df-quickchart-complete');\n",
              "    }\n",
              "    (() => {\n",
              "      let quickchartButtonEl =\n",
              "        document.querySelector('#df-3317b508-e0f2-4b8c-8224-19bdfcedc09f button');\n",
              "      quickchartButtonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "    })();\n",
              "  </script>\n",
              "</div>\n",
              "\n",
              "  <div id=\"id_d0248cca-f9d9-4031-b6f1-226433452011\">\n",
              "    <style>\n",
              "      .colab-df-generate {\n",
              "        background-color: #E8F0FE;\n",
              "        border: none;\n",
              "        border-radius: 50%;\n",
              "        cursor: pointer;\n",
              "        display: none;\n",
              "        fill: #1967D2;\n",
              "        height: 32px;\n",
              "        padding: 0 0 0 0;\n",
              "        width: 32px;\n",
              "      }\n",
              "\n",
              "      .colab-df-generate:hover {\n",
              "        background-color: #E2EBFA;\n",
              "        box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "        fill: #174EA6;\n",
              "      }\n",
              "\n",
              "      [theme=dark] .colab-df-generate {\n",
              "        background-color: #3B4455;\n",
              "        fill: #D2E3FC;\n",
              "      }\n",
              "\n",
              "      [theme=dark] .colab-df-generate:hover {\n",
              "        background-color: #434B5C;\n",
              "        box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "        filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "        fill: #FFFFFF;\n",
              "      }\n",
              "    </style>\n",
              "    <button class=\"colab-df-generate\" onclick=\"generateWithVariable('anthropic_small_sample_df')\"\n",
              "            title=\"Generate code using this dataframe.\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M7,19H8.4L18.45,9,17,7.55,7,17.6ZM5,21V16.75L18.45,3.32a2,2,0,0,1,2.83,0l1.4,1.43a1.91,1.91,0,0,1,.58,1.4,1.91,1.91,0,0,1-.58,1.4L9.25,21ZM18.45,9,17,7.55Zm-12,3A5.31,5.31,0,0,0,4.9,8.1,5.31,5.31,0,0,0,1,6.5,5.31,5.31,0,0,0,4.9,4.9,5.31,5.31,0,0,0,6.5,1,5.31,5.31,0,0,0,8.1,4.9,5.31,5.31,0,0,0,12,6.5,5.46,5.46,0,0,0,6.5,12Z\"/>\n",
              "  </svg>\n",
              "    </button>\n",
              "    <script>\n",
              "      (() => {\n",
              "      const buttonEl =\n",
              "        document.querySelector('#id_d0248cca-f9d9-4031-b6f1-226433452011 button.colab-df-generate');\n",
              "      buttonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "      buttonEl.onclick = () => {\n",
              "        google.colab.notebook.generateWithVariable('anthropic_small_sample_df');\n",
              "      }\n",
              "      })();\n",
              "    </script>\n",
              "  </div>\n",
              "\n",
              "    </div>\n",
              "  </div>\n"
            ],
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "dataframe",
              "variable_name": "anthropic_small_sample_df",
              "summary": "{\n  \"name\": \"anthropic_small_sample_df\",\n  \"rows\": 2,\n  \"fields\": [\n    {\n      \"column\": \"RecordID\",\n      \"properties\": {\n        \"dtype\": \"number\",\n        \"std\": 127786,\n        \"min\": 274046,\n        \"max\": 454763,\n        \"num_unique_values\": 2,\n        \"samples\": [\n          454763,\n          274046\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    },\n    {\n      \"column\": \"Full job description\",\n      \"properties\": {\n        \"dtype\": \"string\",\n        \"num_unique_values\": 2,\n        \"samples\": [\n          \"FEQ225R94\\n\\nThis role can be remote, targeting candidates on the West Coast or can work in the Pacific Time Zone as needed.\\n\\nAs a Specialist Solutions Architect (SSA) - Data Science & Machine Learning, you will be the trusted technical ML expert to both Databricks customers and the Field Engineering organization. You will work with Solution Architects to guide customers in architecting production-grade ML applications on Databricks, while aligning their technical roadmap with the evolving Databricks Data Intelligence Platform. You will continue to strengthen your technical skills through applying the latest technologies in GenAI, MLOps, and ML, while expanding your impact through mentorship and establishing yourself as a ML expert.\\n\\nThe impact you will have:\\n\\nArchitect production level ML workloads for customers using our unified platform, including end-to-end ML pipelines, training, inference optimization, integration with cloud-native services and MLOps\\nProvide advanced technical support to Solution Architects during the technical sale ranging from feature engineering, training, tracking, serving to model monitoring all within a single platform, and participating in the larger ML SME community in Databricks\\nCollaborate with the product and engineering teams to represent the voice of the customer, define priorities and influence the product roadmap, helping with the adoption of Databricks\\u2019 ML offerings\\nBuild and increase customer data science workloads and apply the best MLOps to productionize these workloads across a variety of domains\\nServe as the trusted technical advisor for customers developing GenAI solutions, such as RAG architectures on enterprise knowledge repos, querying structured data with natural language, content generation, and monitoring\\n\\nWhat we look for:\\n\\n5+ years of hands-on industry ML experience in at least one of the following:\\nML Engineer: Develop production-grade cloud (AWS, Azure, GCP) infrastructure that supports the deployment of ML applications, including drift monitoring\\nData Scientist: Experience with the latest techniques in natural language processing including vector databases, fine-tuning LLMs, and deploying LLMs with tools such as HuggingFace, Langchain, and OpenAI\\nGraduate degree in a quantitative discipline (Computer Science, Engineering, Statistics, Operations Research, etc.) or equivalent practical experience\\nExperience communicating and teaching technical concepts to non-technical and technical audiences alike\\nPassion for collaboration, life-long learning, and driving our values through ML\\n[Preferred] 2+ years customer-facing experience in a pre-sales or post-sales role\\n[Preferred] Experience working with Apache Spark\\u2122 to process large-scale distributed datasets\\nCan meet expectations for technical training and role-specific outcomes within 3 months of hire\\nCan travel up to 30% when needed\\n\\nBenefits\\n\\nMedical, Dental, and Vision\\n401(k) Plan\\nFSA, HSA and Commuter Benefit Plans\\nEquity Awards\\nFlexible Time Off\\nPaid Parental Leave\\nFamily Planning\\nFitness Reimbursement\\nAnnual Career Development Fund\\nHome Office, Work Headphones Reimbursement\\nEmployee Assistance Program (EAP)\\nBusiness Travel Accident Insurance\\nMental Wellness Resources\\n\\nPay Range Transparency\\n\\nDatabricks is committed to fair and equitable compensation practices. The pay range(s) for this role is listed below and represents base salary range for non-commissionable roles or on-target earnings for commissionable roles. Actual compensation packages are based on several factors that are unique to each candidate, including but not limited to job-related skills, depth of experience, relevant certifications and training, and specific work location. Based on the factors above, Databricks utilizes the full width of the range. The total compensation package for this position may also include eligibility for annual performance bonus, equity, and the benefits listed above. For more information regarding which range your location is in visit our page here.\\n\\nZone 1 Pay Range\\n$139,800\\u2014$247,300 USD\\nZone 2 Pay Range\\n$139,800\\u2014$247,300 USD\\nZone 3 Pay Range\\n$139,800\\u2014$247,300 USD\\nZone 4 Pay Range\\n$139,800\\u2014$247,300 USD\\nAbout Databricks\\n\\nDatabricks is the data and AI company. More than 10,000 organizations worldwide \\u2014 including Comcast, Cond\\u00e9 Nast, Grammarly, and over 50% of the Fortune 500 \\u2014 rely on the Databricks Data Intelligence Platform to unify and democratize data, analytics and AI. Databricks is headquartered in San Francisco, with offices around the globe and was founded by the original creators of Lakehouse, Apache Spark\\u2122, Delta Lake and MLflow. To learn more, follow Databricks on Twitter, LinkedIn and Facebook.\\n\\nOur Commitment to Diversity and Inclusion\\n\\nAt Databricks, we are committed to fostering a diverse and inclusive culture where everyone can excel. We take great care to ensure that our hiring practices are inclusive and meet equal employment opportunity standards. Individuals looking for employment at Databricks are considered without regard to age, color, disability, ethnicity, family or marital status, gender identity or expression, language, national origin, physical and mental ability, political affiliation, race, religion, sexual orientation, socio-economic status, veteran status, and other protected characteristics.\\n\\nCompliance\\n\\nIf access to export-controlled technology or source code is required for performance of job duties, it is within Employer's discretion whether to apply for a U.S. government license for such positions, and Employer may decline to proceed with an applicant on this basis alone.\",\n          \"Please return soon to complete your dream job application! Senior Principal Machine Learning Engineer Req id: 39356 Virtual, US OPENTEXT OpenText is a global leader in information management, where innovation, creativity, and collaboration are the key components of our corporate culture. As a member of our team, you will have the opportunity to partner with the most highly regarded companies in the world, tackle complex issues, and contribute to projects that shape the future of digital transformation. The Opportunity: As a Machine Learning Engineer at OpenText, you'll embark on an exhilarating journey, facilitating and creating groundbreaking machine learning models and algorithms and developing software products. Your mission? To unravel intricate business puzzles and craft tailored, one-of-a-kind solutions for our valued clientele. Your canvas? A dynamic lab and sandbox environment where innovation knows no bounds. Collaboration is key as you join forces with diverse teams, weaving machine learning and generative AI magic into the fabric of our enterprise solutions. Your creations will not only meet but exceed the most exacting standards of performance, reliability, and scalability. But this role offers more than just challenges; it's an invitation to help shape the future. By contributing to cutting edge technology, your work will leave an indelible mark on global businesses. Welcome to the forefront of innovation, where the intersection of curiosity, creativity, and technology sparks limitless possibilities. You Are Great At: Collaborate with product managers, software engineers, and data scientists to understand business requirements and translate them into machine learning solutions. Design, develop, and deploy machine learning models and algorithms that enhance the functionality and intelligence of our enterprise software products. Explore and analyze large datasets to identify patterns, trends, and insights that can inform the development of robust machine learning models. Implement and optimize machine learning pipelines for training, testing, and deployment of models in production environments. Work on feature engineering, data preprocessing, and model evaluation to ensure the accuracy and effectiveness of machine learning models. Stay abreast of the latest advancements in machine learning and artificial intelligence, and apply relevant techniques to solve real-world problems. Collaborate with cross-functional teams to integrate machine learning capabilities seamlessly into existing and new software products. Conduct code reviews, provide constructive feedback, and contribute to the continuous improvement of development processes. Maintain documentation for machine learning models, algorithms, and implementation details to ensure transparency and knowledge sharing within the team. Stay informed about industry best practices, emerging trends, and advancements in machine learning, and actively contribute to the company's knowledge base. What It Takes: Required: Hold a Master\\u00e2\\u20ac\\u2122s degree or Ph.D. in Computer Science, Applied Mathematics, Statistics, or a related field with a strong emphasis on Natural Language Processing (NLP) and proficiency in Large Language Models (LLMs). Accumulate 3+ years of professional experience, including a significant time dedicated to relevant and hands-on work in Machine Learning, with a specific focus on NLP and LLMs. Proven track record in developing and deploying advanced Deep Learning models, particularly those tailored for NLP tasks and Large Language Models. Demonstrate substantial knowledge and practical experience in implementing Deep Learning models for extracting insights from textual data, Natural Language Understanding, Text Classification, and Document Analysis, all within the context of NLP and LLMs. Possess excellent communication skills, encompassing both verbal and written proficiency, to effectively convey complex concepts and findings related to NLP and Large Language Models. Proven expertise in utilizing cloud-based environments such as AWS, Azure, or GCP Desired: Solid foundational understanding of machine learning and statistical concepts, with a focus on Natural Language Processing (NLP) and Large Language Models (LLMs). Proficiency in deep learning frameworks commonly used in NLP tasks, such as TensorFlow and PyTorch. Hands-on experience in at least two projects involving Large Language Models and other deep learning applications. A minimum of 5+ years of expertise in Convolutional Neural Networks (CNN) and Recurrent Neural Networks (RNN), particularly in the context of NLP, with the ability to develop custom models for various deep learning tasks. At least 5+ years of practical experience working with popular deep learning software frameworks like TensorFlow and PyTorch. Strong programming skills and a background in building deep learning models using one or more of the following languages: Python, R, Java, C, C++, Scala. Position Type and Expected Hours of Work: Full-time, with occasional evenings or weekends based on client needs. Compensation : This job is expected to pay in the range of $230,000-$250,000 USD in addition to a comprehensive and competitive group benefit and healthcare plan. Individual compensation will be determined based on skills and experience comparable to the job requirements. OpenText's commitment to diversity and inclusion surpasses legal requirements, evident in our Equal Employment Opportunity Statement of Policy which promotes a respectful and empowering environment for employees of all backgrounds, culture, national origin, race, color, gender, gender identification, sexual orientation, family status, age, veteran status, disability, religion, or other basis protected by applicable laws. If you need assistance and, or a reasonable accommodation due to a disability during the application or recruiting process, please contact us at hr@opentext.com. Our proactive approach fosters collaboration, innovation, and personal growth, enriching OpenText's vibrant workplace.\"\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    },\n    {\n      \"column\": \"Extracted Technologies\",\n      \"properties\": {\n        \"dtype\": \"string\",\n        \"num_unique_values\": 2,\n        \"samples\": [\n          \"Apache Spark, Delta Lake, MLflow, Databricks, HuggingFace, Langchain, OpenAI\",\n          \"Kubernetes, Docker, MongoDB, PostgreSQL, BigQuery, Spark, Dataflow, JAX, Vertex AI\"\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    }\n  ]\n}"
            }
          },
          "metadata": {},
          "execution_count": 31
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Now let's categorize these\n",
        "def categorize_technologies(technologies):\n",
        "    \"\"\"categorize each of the technologies into one of the\n",
        "    following categories\"\"\"\n",
        "    message = anthropic_client.messages.create(\n",
        "        model=anthropic_model,\n",
        "        max_tokens=1024,\n",
        "        messages=[\n",
        "            {\n",
        "                \"role\": \"user\",\n",
        "                \"content\": cat_prompt + technologies\n",
        "            },\n",
        "        ]\n",
        "    ).content[0].text\n",
        "    return message\n"
      ],
      "metadata": {
        "id": "_iUAMvlArOo5"
      },
      "execution_count": 32,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# Now we parse categorized output\n",
        "def parse_categorized_output(categorized_output):\n",
        "    \"\"\"Parse categorized output\"\"\"\n",
        "    tech_list = categorized_output.split('\\n')\n",
        "    parsed_tech = []\n",
        "\n",
        "    for tech in tech_list:\n",
        "        if ': ' in tech:\n",
        "            technology, category = tech.split(': ')\n",
        "            parsed_tech.append({'Technology': technology, 'Layer': category})\n",
        "\n",
        "    return parsed_tech\n"
      ],
      "metadata": {
        "id": "-FfBpuWPvkxM"
      },
      "execution_count": 33,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# Now we categorize our technologies\n",
        "anthropic_small_sample_df['Categorized Technologies'] = anthropic_small_sample_df['Extracted Technologies'].apply(categorize_technologies)\n"
      ],
      "metadata": {
        "id": "odeH6lijxnDJ"
      },
      "execution_count": 34,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# Finally, we parse all of this and put it in a new df\n",
        "categorized_tech_list = []\n",
        "\n",
        "for _, row in anthropic_small_sample_df.iterrows():\n",
        "    categorized_output = row['Categorized Technologies']\n",
        "    parsed_output = parse_categorized_output(categorized_output)\n",
        "    for entry in parsed_output:\n",
        "        entry['RecordID'] = row['RecordID']\n",
        "        categorized_tech_list.append(entry)\n",
        "\n",
        "categorized_tech_df = pd.DataFrame(categorized_tech_list)\n",
        "\n",
        "# Integrate the categorized technologies back with the original DataFrame\n",
        "final_df = df.merge(categorized_tech_df, on='RecordID', how='left')\n"
      ],
      "metadata": {
        "id": "8ILQXZOxvvT4"
      },
      "execution_count": 35,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "final_df.columns"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "4jJecyV1zTOb",
        "outputId": "fb647f61-5152-4233-8fcf-e3d0840eeab0"
      },
      "execution_count": 36,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Index(['RecordID', 'Job title', 'Full job description', 'Date collected',\n",
              "       'Date posted', 'Week', 'Year', 'Month', 'Quarter', 'Hiring entity',\n",
              "       'Entity size', 'Home office hybrid work', 'Full or part time status',\n",
              "       'Number of applicants', 'Longitude', 'Latitude', 'City', 'Territory',\n",
              "       'Region', 'Experience level', 'Education level', 'Salary start',\n",
              "       'Salary end', 'Main industry', 'Job title category', 'AI technologies',\n",
              "       'AI skills', 'Source name', 'Source URL', 'File Date', 'Technology',\n",
              "       'Layer'],\n",
              "      dtype='object')"
            ]
          },
          "metadata": {},
          "execution_count": 36
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "final_df.shape"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "4Mo5ZRKG1j6I",
        "outputId": "dd0b6b7c-7d11-420c-9ac2-e68fb874cd8c"
      },
      "execution_count": 37,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(205531, 32)"
            ]
          },
          "metadata": {},
          "execution_count": 37
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# prompt: find and remove rows where Technology and Layer are both NaN values\n",
        "final_df_deduped = final_df.dropna(subset=['Technology', 'Layer'])\n"
      ],
      "metadata": {
        "id": "I_KAQ48l1nRp"
      },
      "execution_count": 38,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "final_df_deduped.shape"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "sqqsj1HX1uEk",
        "outputId": "80f08fc9-2dd2-4083-a16e-6b4c42132117"
      },
      "execution_count": 39,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(16, 32)"
            ]
          },
          "metadata": {},
          "execution_count": 39
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "final_df['Layer'].value_counts()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "g5lVylbfysJd",
        "outputId": "d9684f90-9c97-4ce8-c281-5320cea62bfb"
      },
      "execution_count": 40,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Layer\n",
              "Data Layer                       5\n",
              "Infrastructure Layer             3\n",
              "Model Training Layer             3\n",
              "Model Serving Layer              3\n",
              "Pipeline and Automation Layer    2\n",
              "Name: count, dtype: int64"
            ]
          },
          "metadata": {},
          "execution_count": 40
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "final_df['Technology'].value_counts()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "4zHzneSYx-6a",
        "outputId": "ce298efb-19e2-4fbf-b379-a9ad3e2582d3"
      },
      "execution_count": 41,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Technology\n",
              "Kubernetes      1\n",
              "Docker          1\n",
              "MongoDB         1\n",
              "PostgreSQL      1\n",
              "BigQuery        1\n",
              "Spark           1\n",
              "Dataflow        1\n",
              "JAX             1\n",
              "Vertex AI       1\n",
              "Apache Spark    1\n",
              "Delta Lake      1\n",
              "MLflow          1\n",
              "Databricks      1\n",
              "HuggingFace     1\n",
              "Langchain       1\n",
              "OpenAI          1\n",
              "Name: count, dtype: int64"
            ]
          },
          "metadata": {},
          "execution_count": 41
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Create a crosstab for Layer and Technology\n",
        "crosstab = pd.crosstab(final_df['Layer'], final_df['Technology'])\n",
        "crosstab"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 250
        },
        "id": "wfnzrsDYG7vp",
        "outputId": "4f47f542-5a58-4248-cbdd-12ac098573b4"
      },
      "execution_count": 53,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Technology                     Apache Spark  BigQuery  Databricks  Dataflow  \\\n",
              "Layer                                                                         \n",
              "Data Layer                                1         1           0         0   \n",
              "Infrastructure Layer                      0         0           1         0   \n",
              "Model Serving Layer                       0         0           0         0   \n",
              "Model Training Layer                      0         0           0         0   \n",
              "Pipeline and Automation Layer             0         0           0         1   \n",
              "\n",
              "Technology                     Delta Lake  Docker  HuggingFace  JAX  \\\n",
              "Layer                                                                 \n",
              "Data Layer                              1       0            0    0   \n",
              "Infrastructure Layer                    0       1            0    0   \n",
              "Model Serving Layer                     0       0            0    0   \n",
              "Model Training Layer                    0       0            1    1   \n",
              "Pipeline and Automation Layer           0       0            0    0   \n",
              "\n",
              "Technology                     Kubernetes  Langchain  MLflow  MongoDB  OpenAI  \\\n",
              "Layer                                                                           \n",
              "Data Layer                              0          0       0        1       0   \n",
              "Infrastructure Layer                    1          0       0        0       0   \n",
              "Model Serving Layer                     0          1       0        0       1   \n",
              "Model Training Layer                    0          0       0        0       0   \n",
              "Pipeline and Automation Layer           0          0       1        0       0   \n",
              "\n",
              "Technology                     PostgreSQL  Spark  Vertex AI  \n",
              "Layer                                                        \n",
              "Data Layer                              1      0          0  \n",
              "Infrastructure Layer                    0      0          0  \n",
              "Model Serving Layer                     0      0          1  \n",
              "Model Training Layer                    0      1          0  \n",
              "Pipeline and Automation Layer           0      0          0  "
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-aa00c5f2-6224-4d3a-8094-4b4d2f5e7829\" class=\"colab-df-container\">\n",
              "    <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th>Technology</th>\n",
              "      <th>Apache Spark</th>\n",
              "      <th>BigQuery</th>\n",
              "      <th>Databricks</th>\n",
              "      <th>Dataflow</th>\n",
              "      <th>Delta Lake</th>\n",
              "      <th>Docker</th>\n",
              "      <th>HuggingFace</th>\n",
              "      <th>JAX</th>\n",
              "      <th>Kubernetes</th>\n",
              "      <th>Langchain</th>\n",
              "      <th>MLflow</th>\n",
              "      <th>MongoDB</th>\n",
              "      <th>OpenAI</th>\n",
              "      <th>PostgreSQL</th>\n",
              "      <th>Spark</th>\n",
              "      <th>Vertex AI</th>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Layer</th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>Data Layer</th>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Infrastructure Layer</th>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Model Serving Layer</th>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Model Training Layer</th>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Pipeline and Automation Layer</th>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "    <div class=\"colab-df-buttons\">\n",
              "\n",
              "  <div class=\"colab-df-container\">\n",
              "    <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-aa00c5f2-6224-4d3a-8094-4b4d2f5e7829')\"\n",
              "            title=\"Convert this dataframe to an interactive table.\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\" viewBox=\"0 -960 960 960\">\n",
              "    <path d=\"M120-120v-720h720v720H120Zm60-500h600v-160H180v160Zm220 220h160v-160H400v160Zm0 220h160v-160H400v160ZM180-400h160v-160H180v160Zm440 0h160v-160H620v160ZM180-180h160v-160H180v160Zm440 0h160v-160H620v160Z\"/>\n",
              "  </svg>\n",
              "    </button>\n",
              "\n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    .colab-df-buttons div {\n",
              "      margin-bottom: 4px;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "    <script>\n",
              "      const buttonEl =\n",
              "        document.querySelector('#df-aa00c5f2-6224-4d3a-8094-4b4d2f5e7829 button.colab-df-convert');\n",
              "      buttonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "      async function convertToInteractive(key) {\n",
              "        const element = document.querySelector('#df-aa00c5f2-6224-4d3a-8094-4b4d2f5e7829');\n",
              "        const dataTable =\n",
              "          await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                    [key], {});\n",
              "        if (!dataTable) return;\n",
              "\n",
              "        const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "          '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "          + ' to learn more about interactive tables.';\n",
              "        element.innerHTML = '';\n",
              "        dataTable['output_type'] = 'display_data';\n",
              "        await google.colab.output.renderOutput(dataTable, element);\n",
              "        const docLink = document.createElement('div');\n",
              "        docLink.innerHTML = docLinkHtml;\n",
              "        element.appendChild(docLink);\n",
              "      }\n",
              "    </script>\n",
              "  </div>\n",
              "\n",
              "\n",
              "<div id=\"df-d2b3f2bb-9c6d-4948-9da8-f5d1f5d4c47a\">\n",
              "  <button class=\"colab-df-quickchart\" onclick=\"quickchart('df-d2b3f2bb-9c6d-4948-9da8-f5d1f5d4c47a')\"\n",
              "            title=\"Suggest charts\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "<svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "     width=\"24px\">\n",
              "    <g>\n",
              "        <path d=\"M19 3H5c-1.1 0-2 .9-2 2v14c0 1.1.9 2 2 2h14c1.1 0 2-.9 2-2V5c0-1.1-.9-2-2-2zM9 17H7v-7h2v7zm4 0h-2V7h2v10zm4 0h-2v-4h2v4z\"/>\n",
              "    </g>\n",
              "</svg>\n",
              "  </button>\n",
              "\n",
              "<style>\n",
              "  .colab-df-quickchart {\n",
              "      --bg-color: #E8F0FE;\n",
              "      --fill-color: #1967D2;\n",
              "      --hover-bg-color: #E2EBFA;\n",
              "      --hover-fill-color: #174EA6;\n",
              "      --disabled-fill-color: #AAA;\n",
              "      --disabled-bg-color: #DDD;\n",
              "  }\n",
              "\n",
              "  [theme=dark] .colab-df-quickchart {\n",
              "      --bg-color: #3B4455;\n",
              "      --fill-color: #D2E3FC;\n",
              "      --hover-bg-color: #434B5C;\n",
              "      --hover-fill-color: #FFFFFF;\n",
              "      --disabled-bg-color: #3B4455;\n",
              "      --disabled-fill-color: #666;\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart {\n",
              "    background-color: var(--bg-color);\n",
              "    border: none;\n",
              "    border-radius: 50%;\n",
              "    cursor: pointer;\n",
              "    display: none;\n",
              "    fill: var(--fill-color);\n",
              "    height: 32px;\n",
              "    padding: 0;\n",
              "    width: 32px;\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart:hover {\n",
              "    background-color: var(--hover-bg-color);\n",
              "    box-shadow: 0 1px 2px rgba(60, 64, 67, 0.3), 0 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "    fill: var(--button-hover-fill-color);\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart-complete:disabled,\n",
              "  .colab-df-quickchart-complete:disabled:hover {\n",
              "    background-color: var(--disabled-bg-color);\n",
              "    fill: var(--disabled-fill-color);\n",
              "    box-shadow: none;\n",
              "  }\n",
              "\n",
              "  .colab-df-spinner {\n",
              "    border: 2px solid var(--fill-color);\n",
              "    border-color: transparent;\n",
              "    border-bottom-color: var(--fill-color);\n",
              "    animation:\n",
              "      spin 1s steps(1) infinite;\n",
              "  }\n",
              "\n",
              "  @keyframes spin {\n",
              "    0% {\n",
              "      border-color: transparent;\n",
              "      border-bottom-color: var(--fill-color);\n",
              "      border-left-color: var(--fill-color);\n",
              "    }\n",
              "    20% {\n",
              "      border-color: transparent;\n",
              "      border-left-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "    }\n",
              "    30% {\n",
              "      border-color: transparent;\n",
              "      border-left-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "      border-right-color: var(--fill-color);\n",
              "    }\n",
              "    40% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "    }\n",
              "    60% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "    }\n",
              "    80% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "      border-bottom-color: var(--fill-color);\n",
              "    }\n",
              "    90% {\n",
              "      border-color: transparent;\n",
              "      border-bottom-color: var(--fill-color);\n",
              "    }\n",
              "  }\n",
              "</style>\n",
              "\n",
              "  <script>\n",
              "    async function quickchart(key) {\n",
              "      const quickchartButtonEl =\n",
              "        document.querySelector('#' + key + ' button');\n",
              "      quickchartButtonEl.disabled = true;  // To prevent multiple clicks.\n",
              "      quickchartButtonEl.classList.add('colab-df-spinner');\n",
              "      try {\n",
              "        const charts = await google.colab.kernel.invokeFunction(\n",
              "            'suggestCharts', [key], {});\n",
              "      } catch (error) {\n",
              "        console.error('Error during call to suggestCharts:', error);\n",
              "      }\n",
              "      quickchartButtonEl.classList.remove('colab-df-spinner');\n",
              "      quickchartButtonEl.classList.add('colab-df-quickchart-complete');\n",
              "    }\n",
              "    (() => {\n",
              "      let quickchartButtonEl =\n",
              "        document.querySelector('#df-d2b3f2bb-9c6d-4948-9da8-f5d1f5d4c47a button');\n",
              "      quickchartButtonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "    })();\n",
              "  </script>\n",
              "</div>\n",
              "\n",
              "  <div id=\"id_3f39034b-fd52-4c47-b599-336affcd2ede\">\n",
              "    <style>\n",
              "      .colab-df-generate {\n",
              "        background-color: #E8F0FE;\n",
              "        border: none;\n",
              "        border-radius: 50%;\n",
              "        cursor: pointer;\n",
              "        display: none;\n",
              "        fill: #1967D2;\n",
              "        height: 32px;\n",
              "        padding: 0 0 0 0;\n",
              "        width: 32px;\n",
              "      }\n",
              "\n",
              "      .colab-df-generate:hover {\n",
              "        background-color: #E2EBFA;\n",
              "        box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "        fill: #174EA6;\n",
              "      }\n",
              "\n",
              "      [theme=dark] .colab-df-generate {\n",
              "        background-color: #3B4455;\n",
              "        fill: #D2E3FC;\n",
              "      }\n",
              "\n",
              "      [theme=dark] .colab-df-generate:hover {\n",
              "        background-color: #434B5C;\n",
              "        box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "        filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "        fill: #FFFFFF;\n",
              "      }\n",
              "    </style>\n",
              "    <button class=\"colab-df-generate\" onclick=\"generateWithVariable('crosstab')\"\n",
              "            title=\"Generate code using this dataframe.\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M7,19H8.4L18.45,9,17,7.55,7,17.6ZM5,21V16.75L18.45,3.32a2,2,0,0,1,2.83,0l1.4,1.43a1.91,1.91,0,0,1,.58,1.4,1.91,1.91,0,0,1-.58,1.4L9.25,21ZM18.45,9,17,7.55Zm-12,3A5.31,5.31,0,0,0,4.9,8.1,5.31,5.31,0,0,0,1,6.5,5.31,5.31,0,0,0,4.9,4.9,5.31,5.31,0,0,0,6.5,1,5.31,5.31,0,0,0,8.1,4.9,5.31,5.31,0,0,0,12,6.5,5.46,5.46,0,0,0,6.5,12Z\"/>\n",
              "  </svg>\n",
              "    </button>\n",
              "    <script>\n",
              "      (() => {\n",
              "      const buttonEl =\n",
              "        document.querySelector('#id_3f39034b-fd52-4c47-b599-336affcd2ede button.colab-df-generate');\n",
              "      buttonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "      buttonEl.onclick = () => {\n",
              "        google.colab.notebook.generateWithVariable('crosstab');\n",
              "      }\n",
              "      })();\n",
              "    </script>\n",
              "  </div>\n",
              "\n",
              "    </div>\n",
              "  </div>\n"
            ],
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "dataframe",
              "variable_name": "crosstab",
              "summary": "{\n  \"name\": \"crosstab\",\n  \"rows\": 5,\n  \"fields\": [\n    {\n      \"column\": \"Layer\",\n      \"properties\": {\n        \"dtype\": \"string\",\n        \"num_unique_values\": 5,\n        \"samples\": [\n          \"Infrastructure Layer\",\n          \"Pipeline and Automation Layer\",\n          \"Model Serving Layer\"\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    },\n    {\n      \"column\": \"Apache Spark\",\n      \"properties\": {\n        \"dtype\": \"number\",\n        \"std\": 0,\n        \"min\": 0,\n        \"max\": 1,\n        \"num_unique_values\": 2,\n        \"samples\": [\n          0,\n          1\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    },\n    {\n      \"column\": \"BigQuery\",\n      \"properties\": {\n        \"dtype\": \"number\",\n        \"std\": 0,\n        \"min\": 0,\n        \"max\": 1,\n        \"num_unique_values\": 2,\n        \"samples\": [\n          0,\n          1\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    },\n    {\n      \"column\": \"Databricks\",\n      \"properties\": {\n        \"dtype\": \"number\",\n        \"std\": 0,\n        \"min\": 0,\n        \"max\": 1,\n        \"num_unique_values\": 2,\n        \"samples\": [\n          1,\n          0\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    },\n    {\n      \"column\": \"Dataflow\",\n      \"properties\": {\n        \"dtype\": \"number\",\n        \"std\": 0,\n        \"min\": 0,\n        \"max\": 1,\n        \"num_unique_values\": 2,\n        \"samples\": [\n          1,\n          0\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    },\n    {\n      \"column\": \"Delta Lake\",\n      \"properties\": {\n        \"dtype\": \"number\",\n        \"std\": 0,\n        \"min\": 0,\n        \"max\": 1,\n        \"num_unique_values\": 2,\n        \"samples\": [\n          0,\n          1\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    },\n    {\n      \"column\": \"Docker\",\n      \"properties\": {\n        \"dtype\": \"number\",\n        \"std\": 0,\n        \"min\": 0,\n        \"max\": 1,\n        \"num_unique_values\": 2,\n        \"samples\": [\n          1,\n          0\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    },\n    {\n      \"column\": \"HuggingFace\",\n      \"properties\": {\n        \"dtype\": \"number\",\n        \"std\": 0,\n        \"min\": 0,\n        \"max\": 1,\n        \"num_unique_values\": 2,\n        \"samples\": [\n          1,\n          0\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    },\n    {\n      \"column\": \"JAX\",\n      \"properties\": {\n        \"dtype\": \"number\",\n        \"std\": 0,\n        \"min\": 0,\n        \"max\": 1,\n        \"num_unique_values\": 2,\n        \"samples\": [\n          1,\n          0\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    },\n    {\n      \"column\": \"Kubernetes\",\n      \"properties\": {\n        \"dtype\": \"number\",\n        \"std\": 0,\n        \"min\": 0,\n        \"max\": 1,\n        \"num_unique_values\": 2,\n        \"samples\": [\n          1,\n          0\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    },\n    {\n      \"column\": \"Langchain\",\n      \"properties\": {\n        \"dtype\": \"number\",\n        \"std\": 0,\n        \"min\": 0,\n        \"max\": 1,\n        \"num_unique_values\": 2,\n        \"samples\": [\n          1,\n          0\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    },\n    {\n      \"column\": \"MLflow\",\n      \"properties\": {\n        \"dtype\": \"number\",\n        \"std\": 0,\n        \"min\": 0,\n        \"max\": 1,\n        \"num_unique_values\": 2,\n        \"samples\": [\n          1,\n          0\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    },\n    {\n      \"column\": \"MongoDB\",\n      \"properties\": {\n        \"dtype\": \"number\",\n        \"std\": 0,\n        \"min\": 0,\n        \"max\": 1,\n        \"num_unique_values\": 2,\n        \"samples\": [\n          0,\n          1\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    },\n    {\n      \"column\": \"OpenAI\",\n      \"properties\": {\n        \"dtype\": \"number\",\n        \"std\": 0,\n        \"min\": 0,\n        \"max\": 1,\n        \"num_unique_values\": 2,\n        \"samples\": [\n          1,\n          0\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    },\n    {\n      \"column\": \"PostgreSQL\",\n      \"properties\": {\n        \"dtype\": \"number\",\n        \"std\": 0,\n        \"min\": 0,\n        \"max\": 1,\n        \"num_unique_values\": 2,\n        \"samples\": [\n          0,\n          1\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    },\n    {\n      \"column\": \"Spark\",\n      \"properties\": {\n        \"dtype\": \"number\",\n        \"std\": 0,\n        \"min\": 0,\n        \"max\": 1,\n        \"num_unique_values\": 2,\n        \"samples\": [\n          1,\n          0\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    },\n    {\n      \"column\": \"Vertex AI\",\n      \"properties\": {\n        \"dtype\": \"number\",\n        \"std\": 0,\n        \"min\": 0,\n        \"max\": 1,\n        \"num_unique_values\": 2,\n        \"samples\": [\n          1,\n          0\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    }\n  ]\n}"
            }
          },
          "metadata": {},
          "execution_count": 53
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# prompt: Create a chart that shows the correlation between Layer and Technology\n",
        "correlation_matrix = final_df_deduped.groupby(['Layer', 'Technology']).size().unstack()\n",
        "\n",
        "# Create a heatmap of the correlation matrix\n",
        "plt.figure(figsize=(5, 5))\n",
        "sns.heatmap(correlation_matrix, annot=True, cmap='coolwarm')\n",
        "plt.xlabel('Technology')\n",
        "plt.ylabel('Layer')\n",
        "plt.title('Correlation between Layer and Technology')\n",
        "plt.show()\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 569
        },
        "id": "FQLcdJlQ5XXW",
        "outputId": "6f8dce30-984a-40f8-8914-cb4c15651edb"
      },
      "execution_count": 42,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 500x500 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    }
  ]
}