During January 2025, Fillzpp enhanced the GPU monitoring test infrastructure for the leptonai/gpud repository by refactoring end-to-end tests to use mock implementations of NVML, nvidia-smi, and lspci. This approach reduced test flakiness and improved the reliability of continuous integration pipelines. Working primarily with Go and Bash, Fillzpp focused on system integration and mocking techniques to create deterministic and robust test scenarios for GPU-related features. Additionally, they updated the Codecov configuration to exclude the e2e directory, ensuring more accurate coverage metrics. The work demonstrated a thoughtful approach to test reliability and maintainability within a complex system environment.