Kfops

Introduction

Read the high-level introductory article.

Getting started & article tldr;

Kfops is an opinionated Kubeflow "wrapper" to manage your pipelines and model deployment with the use of dedicated python package and chatops commands. It simplifies and standardizes the Kubeflow-based ML model lifecycle by enforcing the "single repository per ML model" rule and the explicit way how it should be configured. Commands like /build, /build_run and /deploy are executed in the context of the Pull Request.

Notice: Kfops requires "full" Kubeflow deployment (version >= 1.3) and KServe to serve models. Check other requirements in the Administrator's guide.

For details, check:

• Commands reference
• Where does Kfops fit in the MLOps lifecycle
• Design goals and features

Installation steps

Refer to Installation steps for details.