Worked on enhancing continuous integration reliability for the ROCm/TheRock repository by introducing automated retries and timeout monitoring to GPU test workflows. Leveraged GitHub Actions and YAML to implement prolense-enabled retries, using environment variables to stabilize flaky GPU runs and reduce CI noise. Added comprehensive timeout controls and standardized metadata for improved dashboard visibility, enabling more consistent and actionable test reporting. Focused on end-to-end workflow improvements across test_component and setup_test_environment actions, accelerating feedback on GPU-related changes. The work emphasized robust CI/CD practices, workflow automation, and testing, resulting in more reliable and maintainable GPU test pipelines for the project.