Contributed to the apache/gravitino repository by delivering features and fixes that enhanced backend reliability and data access efficiency. Focused on reducing database round-trips, the work introduced batch metadata name resolution using Java, MyBatis, and SQL, eliminating N+1 query issues for key entities. Addressed data consistency by centralizing schema persistence error handling and implementing automatic rollback to maintain catalog integrity. Improved legacy data cleanup through corrected deletion queries and validated input handling with targeted unit tests. These efforts strengthened error diagnosability and system resilience, demonstrating a methodical approach to backend development, database management, and test-driven quality across relational data stores.