Worked on enhancing distributed training capabilities in the aws/deep-learning-containers repository by introducing the SageMaker Distributed Data Parallel (SMDDP) binary for PyTorch 2.4. This involved updating Dockerfiles and build specifications to ensure seamless installation and configuration of the new binary, enabling scalable training workflows in AWS SageMaker environments. Adjusted the testing logic to accommodate changes in PyTorch 2.4 SMDDP, which improved test coverage for distributed training scenarios. The work was implemented using Python and YAML, leveraging skills in Docker and distributed computing to deliver a robust feature that supports end-to-end compatibility for large-scale machine learning workloads.