Worked on the pytorch/pytorch repository to expand performance testing capabilities by integrating AMD AWS runners into the continuous integration pipeline. This feature enabled AMD-specific performance benchmarks and broadened hardware coverage for the project. The technical approach involved updating CI/CD configurations using Shell and YAML, as well as adding architecture-specific settings to support multiple AMD platforms. Documentation and infrastructure were revised to reflect the new runner usage, ensuring clarity for future contributors. By improving the feedback loop for performance tests on AMD hardware, the work accelerated the identification of regressions and enhanced the reliability of performance testing within the PyTorch CI environment.