Worked on the AI-Hypercomputer/xpk repository to stabilize and enhance DWS workload management, focusing on both risk reduction and feature delivery. Addressed deployment stability by reverting TAS integration and refining configuration logic across multiple Python modules. Introduced new TAS support for DWS clusters, enabling pod set updates with node selectors for dynamic resizing. Improved the robustness of the tcpx_decorator by adding necessary volume mounts and expanded unit test coverage for annotations, tolerations, and environment variables. Leveraged Python and YAML within a Kubernetes and DevOps context, ensuring safer, faster iteration cycles and supporting scalable workload capabilities through targeted bug fixes and refactoring.