Worked on the apache/spark repository to deliver a performance optimization feature targeting large table property reading in Spark’s metadata path. Addressed scalability challenges by reducing the algorithmic complexity from O(N^2) to O(N), modifying the logic to construct properties only when numParts exists. This approach improved query performance and reduced CPU and memory usage during SQL property handling for large tables. The work involved code-level changes and review, leveraging Scala and Java for backend development and algorithmic optimization. No major bugs were fixed during this period, with the primary focus on enhancing efficiency and scalability in Spark’s metadata property management.