Machine Learning at Scale Using KubeFlow on AWS
summary
Shows a reproducible pipeline: creates an Amazon EFS file system, installs the EFS CSI driver and a storage class, mounts an EFS-backed PVC into a Kubeflow Jupyter notebook, builds a Docker image on Cloud9, pushes it to ECR, and runs a TFJob with 2 replicas that trains on a small flowers dataset stored on Amazon EFS. Demo completes quickly and works end-to-end; EFS claims sub-millisecond reads (~600 microseconds) and the notebook + TFJob flow is validated.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 0:49Why Machine Learning with Containers
- 5:15KubeFlow
- 13:26Setup
- 18:34efs
- 20:03Create Volume
- 20:32Create Jupyter Notebook
- 22:30Download Data Set
related talks
54mMachine learning at scale using Kubeflow with Amazon EKS & Amazon EFS
51mMachine Learning using Kubeflow and Kubernetes
46mAccelerating Machine Learning DevOps with Kubeflow
46mMachine Learning with Kubernetes
34mBuilding End-to-End Machine Learning Workflows with Kubernetes...
28mMachine Learning on Kubernetes at Shell: A Kubeflow Journey
46mDeploying scalable Machine Learning models with Kubernetes
18mKeynote: Containerizing Deep Learning Using Kubeflow on AWS