Deploying scalable Machine Learning models with Kubernetes
summary
Concrete how-to for running ML on Kubernetes using Kubeflow components: Jupyter notebooks, TFJob training, Katib hyperparameter search, Pachyderm for data versioning, Argo for GitOps pipelines and model serving. Demonstrates Azure features: Azure Container Instances plus a virtual node overflow that can present ACI with GPU capacity (demo notes a preview break but describes scheduling up to 100 GPU-backed containers), claims portability across clouds, production users (Uber) and a Kubeflow Pipelines dashboard that currently runs on Google Cloud.
AI-written summary · about
chapters (12)
- 0:00Introduction to talk
- 2:31Cloud Native Computing Foundation
- 7:34Machine learning impact and hype
- 13:20Kubernetes overview and design
- 18:50Kubernetes operators explained
- 21:41Microsoft Azure and Kubernetes
- 24:58Introducing Kubeflow
- 27:26Kubeflow components and benefits
- 31:21Challenges in ML deployment
- 33:28Kubeflow pipelines and workflow
- 37:33Demo and Kubeflow summary
- 41:17Q&A and community resources
related talks
54mScaling Machine Learning Pipelines in Cloud
51mMachine Learning using Kubeflow and Kubernetes
31mKubernetes for Machine Learning: Managing and Scaling AI Workloads
39mDeploy your ML Workloads on Kubernetes with Kubeflow
42mKubernetes for Machine Learning: Managing and Scaling AI Workloads
22mThe Containerization of Machine Learning: TensorFlow: Kubernetes, and Kubeflow
54mEconomics and Best Practices of Running AI/ML Workloads on Kubernetes
39mOperating Deep Learning Pipelines Anywhere Using Kubeflow