Amazon EC2 offers the broadest set of accelerators in the cloud to run machine learning workloads. Whether you are using Amazon EC2 instances powered by NVIDIA GPUs AWS Trainium, Inferentia you may wonder how to manage the software stack. What to put in the Amazon Machine Image (AMI)? What goes on the container? How would that be different on Kubernetes with Amazon EKS or on Slurm AWS ParallelCluster? How to make use of the DLAMI. In this chalk-talk we will dive into the software stack for different accelerators, services and containerization systems. Dive into AMIs and techniques to build and manage your software stack from practical experience.