During May 2025, Zshu developed an end-to-end test for the llama_3_1_405b_nemo model targeting a3ultra hardware within the GoogleCloudPlatform/ml-auto-solutions repository. Leveraging Python and MLOps practices, Zshu designed a DAG-based workflow that automates hardware validation and accelerates feedback cycles for model deployment. The work included refining Docker image selection logic to ensure model-specific images are used, reducing misconfiguration risks and deployment delays. Updates to CODEOWNERS reflected expanded test coverage and clarified responsibilities. This feature enhanced CI/CD reliability for hardware-accelerated inference, demonstrating depth in cloud engineering and a focus on safer, more efficient release processes without addressing critical bugs.