Worked on the leptonai/gpud repository to enhance GPU monitoring test infrastructure, focusing on improving the reliability of continuous integration workflows. Refactored end-to-end tests by introducing mock implementations for NVML, nvidia-smi, and lspci, which reduced test flakiness and enabled deterministic test scenarios. Utilized Go and Bash scripting to build and integrate these mocks, supporting more robust system integration and end-to-end testing. Updated the Codecov configuration to exclude the e2e directory, ensuring more accurate coverage metrics and clearer signals of test quality. The work addressed CI noise and contributed to more maintainable and reliable GPU-related feature testing.