Jiya Zhang enhanced reliability and test coverage across two cloud engineering projects over a two-month period. On the apple/axlearn repository, Jiya improved the Goodput measurement system by enforcing single-process initiation for the uploader, reducing duplicate uploads, and delivering clear documentation for AXLearn training jobs. This work, implemented in Python and leveraging GCP, included targeted unit testing to safeguard process management. In the GoogleCloudPlatform/ml-auto-solutions repository, Jiya updated TPU test suites for v5p and v6e compatibility, reconfiguring job parameters and upgrading dependencies to support new hardware. These contributions strengthened workflow stability and ensured robust, up-to-date testing for production ML workloads.