add a Kaggle example

[jameslamb]

Description

Kaggle offers a hosted-notebook environment to run workloads that RAPIDS is great for: exploratory data analysis, machine learning training, and machine learning model evaluation.

That notebook environment has some RAPIDS libraries (like cuDF and cuML) installed by default, and the ability to use NVIDIA GPUs.

The RAPIDS deployment docs should document this and provide an example to help folks get started.

Benefits of this work

• documents how to use RAPIDS for a popular platform that already comes with it some RAPIDS things pre-installed

Acceptance Criteria

• one or more pages exist at https://docs.rapids.ai/deployment/stable/ describing how to use RAPIDS on Kaggle notebooks

Approach

Go to https://www.kaggle.com/code and click + New Notebook to experiment with what's available.

Create a new page documenting how to best run some RAPIDS-y workloads in that environment.

References:

• https://www.kaggle.com/code
• https://www.kaggle.com/docs/efficient-gpu-usage
• docker image defining the software available there: https://github.com/Kaggle/docker-python

Consider involving the Kaggle folks at NVIDIA for additional guidance.

Notes

N/A