Enabling Multi-user Machine Learning Workflows for Kubeflow Pipelines

Explore how to enable multi-user machine learning workflows for Kubeflow Pipelines in this 26-minute conference talk from CNCF. Discover the challenges and solutions for implementing access control, authentication, and authorization in Kubeflow, an open-source machine learning platform built on Kubernetes. Learn about combining cloud-native technologies to create a flexible, Kubernetes-native solution for services with their own API and database. Gain insights into securing in-cluster traffic, implementing centralized API servers, and utilizing decentralized UI artifact servers. Watch a live demo and understand the design and implementation process, including the use of Istio for enhanced security and multi-user support in Kubeflow Pipelines.

Enabling Multi-user Machine Learning Workflows for Kubeflow Pipelines
Auth for Kubeflow Pipelines - Authentication HTTP Headers
Securing in-cluster traffic
Multi user support for KFP
Centralized API Server Pros
Decentralized UI artifact servers
Design - what is missing?
Implementation - istio