Skip to main content

Getting Started

note

What is CAIP Inference?

CAIP Inference API abstracts the complexity of managing AI infrastructure resources in a distributed cloud environment. Built on Ray framework and MLFlow, it exposes independent CRUD operations for two inference modes: Live Inference for real-time model serving, and Batch Inference for asynchronous job processing.

Features

This API currently features both types of inference tasks that can be done by the users:

Live Inference

  • Create a live inference application by specifying MLFlow models
  • List your currently available live inference applications
  • Check a live inference application's status
  • Perform a live inference prediction
  • Delete a live inference application

Batch Inference

  • Create a batch inference job for currently available models
  • List your created and ongoing batch inference jobs
  • Check a batch inference job's status
  • Delete a batch inference job

Quickstart

Get started quickly with these hands-on guides:

Live Inference Usage

Learn how to use the Inference API to execute Live Inference jobs.

Start Tutorial →

Batch Inference Usage

Learn how to use the Inference API to execute Batch Inference jobs.

Start Tutorial →