Kubeflow Deep Dive
summary
Kubeflow presents a Kubernetes-native ML platform built from ksonnet packages and custom resources that already supports TF operator, a forthcoming PyTorch operator, Katib hyperparameter tuning and integrations with Argo, Pachyderm, JupyterHub and Seldon. The project claims about 35 PRs per week and 70 individual members across 20+ organizations, shipped a 0.1 release in April, targets 0.2 this quarter and 1.0 by December with a CI/CD-for-ML goal and principles of openness, low barrier / high ceiling, and declarative ops.
AI-written summary · about
chapters (20)
- 0:00Introduction
- 0:51Agenda
- 1:32The Problem
- 2:54The Goal
- 3:12Kubeflow
- 6:31Community
- 7:50ML Platform
- 8:42ML Development
- 9:31Component Types
- 10:20Projects
- 13:06Road Map
- 13:53Core Components
- 14:25Continuous Integration and Deployment
- 15:30New Components
- 16:16Core Principles
- 16:28Being Open
- 17:56Low Bar High Ceiling
- 19:29Kubeflow Native
- 21:48Kubeflow Low Bar
- 26:43Bootstrapper
related talks
14mKubeflow: Where ML Meets Kubernetes
37mTowards Kubeflow 1.0, Bringing a Cloud Native Platform For ML to Kubernetes
31mKubeflow 2026 State of the Union: Orchestrating the Full ML Lifecycle
21mIntroducing Kubeflow (w. Special Guests Tensorflow and Apache Spark)
32mKubeflow Ecosystem: What’s Next for Cloud Native AI/ML and LLMOps
30mKubeflow, Your ML Tools on Kubernetes
40mKubeflow: ML on OpenShift
36mIntroducing Kubeflow (w. Special Guests Tensorflow and Apache Spark)