Optimized Jupyter Notebooks on AWS

AWS-optimized Notebooks based on AWS Deep Learning Containers

AWS Optimized Notebook Images

Installing Kubeflow on AWS using this guide will include AWS-optimized Kubeflow Notebook Images as the default options in the notebook server.

These images are based upon AWS Deep Learning Containers. AWS Deep Learning Containers provide optimized environments with TensorFlow and MXNet, Nvidia CUDA (for GPU instances), and Intel MKL (for CPU instances) libraries.

Additional pre-installed packages:

  • docker-client
  • kubeflow-metadata
  • kfp
  • kfserving

These images are available from the Amazon Elastic Container Registry (Amazon ECR).