This is the Docker & Kubernetes config that accompanies my blog post Deploying ML workloads with Azure and Kubernetes.

This uses the Evolution Strategies starter and Roboschool.

You should first create the docker container image and test locally..

cd docker
docker build -t gretkowski/kubeevo:v1 -f Dockerfile .

make sure everything is good w/ the container image, then push it to your private image registry.

You then use the rest of the YAML files - modified to suit your environment - to deploy the master and worker containers to your Kubernetes cluster