Skip to main content

KubeDL makes deep learning workloads run on Kubernetes more easily and efficiently.

Various ML Framework Support

Support training and serving workloads in a single controller with enhancements in scheduling, performance, metadata persistency and more.

Model Package, Deployment and Tracking

Package and deploy with container, tracking model lineage natively in Kubernetes CRD.

Auto Tuning for Model Deployment

Automatic tuning of container configurations to maximize runtime and cost efficiency of model serving.


KubeDL is a Cloud Native Computing Foundation sandbox project