Worked on improving the reliability of the testing framework in the trinodb/trino repository by addressing a race condition during asynchronous cleanup. Using Java and asynchronous programming techniques, introduced a retry-based synchronization mechanism that ensures cleanup processes complete before test verifications proceed. This approach targeted issues in high-latency environments such as MinIO, reducing the occurrence of flaky tests and stabilizing continuous integration pipelines. The solution included thorough root-cause analysis and commit-level traceability, enabling reproducibility and easier maintenance. Focused on testing and bug resolution, the work accelerated feedback cycles for feature development and increased confidence in the release process.