Building a MLOps Platform with Kubeflow on GKE
summary
Shows a concrete stack and what each piece actually does: Kubeflow Pipelines for containerized step orchestration, Argo under the hood, TFJob for distributed training, JupyterHub notebooks for prototyping, images pushed to GCR and run on GKE with autoscaling, and KServe plus model artifacts in Cloud Storage to expose live endpoints. Demonstrates hyperparameter search with Katib, cites ~65% of k8s ML workloads, and provides workshop code labs that build and run the platform end to end.
AI-written summary · about
chapters (8)
- 0:00Start
- 2:13Typical machine learning lifecycle Step-by-step solution of ML problem
- 7:27What does Kubeflow Pipelines offer?
- 12:03Kubeflow Pipelines SDK - Anatomy of a
- 13:01Kubeflow Pipelines SDK - Pipeline
- 14:44Training - Steps
- 15:18Move container to GCR
- 16:51Serving
related talks
20mML Ops design patterns with Kubeflow Pipelines
51mMachine Learning using Kubeflow and Kubernetes
46mDeploying scalable Machine Learning models with Kubernetes
01h 00m#MLOps for #Java Developers A Hands On Guide with #Kubeflow and #Quarkus
26mKubernetes and MLOps for Scalable and Reproducible Generative AI
31mKubernetes and MLOps for Scalable and Reproducible Generative AI
44mEngaging the Kubeflow Community: Building an Enterprise-Ready AI/ML Platform
14mKubeflow: Where ML Meets Kubernetes