Worked on the Meesho/BharatMLStack repository to deliver targeted enhancements to the BharatML SDK, focusing on expanding data type support and improving naming clarity within data pipelines. Leveraging Python and data engineering skills, the work introduced compatibility for new data types, aligning with cross-team data contracts and supporting downstream analytics and machine learning workloads. Addressed a parsing issue by correcting a typo in unsigned integer handling, thereby safeguarding data integrity. Enhanced the column renaming process to include feature group labels, which reduced naming conflicts and improved data lineage across feature stores. The contributions emphasized reliability and maintainability in SDK development.