Worked on the ray-project/ray repository to enhance data ingestion stability and CI reliability for enterprise-scale workloads. Developed a row-group-aware Parquet chunking mechanism using Python and PyArrow, aligning data chunks with actual row group boundaries to improve processing throughput and reduce read errors. Improved CI/CD processes by increasing the Iceberg test suite timeout, allowing more complex and longer-running tests to complete successfully. Leveraged multithreading and filesystem-aware chunking to parallelize metadata reads, boosting scalability across large datasets. Demonstrated skills in data processing, file handling, and test infrastructure, delivering targeted improvements that strengthened both data pipeline efficiency and continuous integration robustness.