A Component Registry for Kubeflow Pipelines
Talkat Kubernetes AI Day Europe 2022 · on Monday, 2022-05-16· at 14:40· Pavilion 3, Room D | Level 2 | Central Forum·
summary
Proposal for a centralized component registry API that stores Kubeflow Pipelines components as unified YAML packages with versioning and tags (Docker-like), plus upload/download/list endpoints and SDK hooks to load components by package+tag. A working catalog called Machine Learning Exchange (MLX) bundles that registry with an execution engine (Tekton-backed KFP), integrations for DataSHIM, KServe, Lyra and Model/Data Asset Exchange, a notebook runner using Papermill, and a demoed language classifier from a ~14M sample corpus improving from 8/10 to 9/10 accuracy.
AI-written summary · about
chapters (19)
- 0:00Introduction
- 0:27Agenda
- 3:00Challenges
- 3:53Q4 Pipelines
- 5:42Components
- 7:06Component Ecosystem
- 7:53Existing Kubeflow Ecosystem
- 9:41Component Registry
- 11:00Component Registry Terminology
- 11:23Component Registry Examples
- 12:29Machine Learning Exchange
- 13:06ML Exchange
- 14:29User Interface
- 16:42Component List
- 17:07Integrations
- 18:45Data Sets
- 26:00Template Registry
- 27:45Links
- 28:18Outro
related talks
10mLightning Talk: A Component Registry for Kubeflow Pipelines
35mOrganizing data engineering around Kubeflow
29mML Engineering Kubeflow
31mKubeflow 2026 State of the Union: Orchestrating the Full ML Lifecycle
09mKubeflow Pipelines v2: The Next Generation of MLOps on Ku...
14mKubeflow: Where ML Meets Kubernetes
39mDeploy your ML Workloads on Kubernetes with Kubeflow
18mKubeflow Ecosystem: Past, Present, Futu...