Serve prediction requests using Knative Serving and AWS Load Balancer
Run inference using Kubeflow on AWS using AWS Deep Learning Containers