Worked on stabilizing the test suite for laude-institute/terminal-bench, focusing on improving the reliability of asynchronous test flows. Addressed a persistent flakiness issue by increasing the sleep duration in a Python test case, allowing asynchronous tasks to complete before a SIGINT was sent. This adjustment reduced race conditions and made continuous integration results more deterministic, directly supporting faster and safer pull request validation. The work did not introduce new user-facing features but delivered a targeted bug fix that enhanced CI reliability. Demonstrated proficiency in Python and testing methodologies, with careful attention to timing and reproducibility in automated test environments.