Developed a benchmark configuration and metrics suite for regression testing in the pytorch/test-infra repository, focusing on the integration of PyTorch with vLLM. Leveraged backend development and data analysis skills to define performance thresholds and automate regression detection, linking results directly to GitHub issues for streamlined triage. Utilized Python to implement a repeatable local workflow, incorporating AWS Lambda for reporting and ClickHouse for metrics storage. This work improved visibility into performance regressions and enabled more reliable continuous integration by capturing and evaluating key metrics, ultimately supporting faster identification and resolution of issues within the PyTorch/test-infra ecosystem.