Building ML Products With Kubeflow
summary
Project demonstrates Kubeflow running on Kubernetes with a concrete GitHub-issue-title autogeneration product: interactive development in JupyterHub on curated containers, quick proof-of-concept on a downsampled archive, distributed training via the TFJob operator to scale beyond the notebook, metrics in TensorBoard, deployment as a microservice using Seldon model server, HTTP frontend and ingress via Ambassador, on-prem GPU support and a conjure-up installer; community has ~2200 GitHub stars and multi-vendor contributors.
AI-written summary · about
chapters (19)
- 0:00Intro
- 0:21Agenda
- 1:10Kubeflow
- 1:49Machine Learning Everywhere
- 2:54Machine Learning Code
- 4:03DevOps
- 4:19Kubeflow Project
- 4:53Kubeflow Overview
- 7:41Deploy Kubeflow
- 8:17Deploy Canonical
- 9:02Kubeflow Workflow
- 9:59Identifying a Problem
- 10:57GitHub Issues
- 11:16Jupiter Notebooks
- 15:28Training
- 20:41Web App
- 21:27Demo Summary
- 23:23Roadmap
- 24:55Summary
related talks
37mTowards Kubeflow 1.0, Bringing a Cloud Native Platform For ML to Kubernetes
14mKubeflow: Where ML Meets Kubernetes
46mDeploying scalable Machine Learning models with Kubernetes
51mMachine Learning using Kubeflow and Kubernetes
39mDeploy your ML Workloads on Kubernetes with Kubeflow
31mKubeflow - Data Science on Steroids
21mIntroducing Kubeflow (w. Special Guests Tensorflow and Apache Spark)
22mBuilding a MLOps Platform with Kubeflow on GKE