Skip to content

Latest commit

 

History

History
25 lines (22 loc) · 949 Bytes

ROADMAP.md

File metadata and controls

25 lines (22 loc) · 949 Bytes

KServe Roadmap

2021 Q4/2022 Q1

Kubernetes Deployment

Objective: "Enable raw kubernetes deployment as alternative mode"

  • Support existing ML frameworks, transformer/explainer, logger and batching
  • Make Istio/KNative optional and unlock KNative limitations
    • Allow multiple volumes mounted
    • Allow TCP/UDP

Inference Graph

Objective: "Enable model serving pipelines with flexible routing graph"

ModelMesh

Objective: "Unifying interface for SingleModel and ModelMesh deployment"

  • Ability to perform inference using Predict v2 API with REST/gRPC
  • Unify the storage support for single and ModelMesh
  • InferenceService controller to utilize ServingRuntime
  • Single install for KServe which includes SingleModel and ModelMesh Serving