Worked on the apache/spark repository to address a critical correctness issue in the Parquet vectorized reader, specifically targeting the handling of nested arrays that span multiple pages. Using Java and Scala, applied expertise in Apache Spark, big data, and data processing to correct row index usage during the explode operation, ensuring accurate processing of complex nested Parquet data. Developed and integrated regression tests to validate the fix and reinforce coverage for edge-case nested structures. This work improved data correctness and reduced the risk of data corruption for users processing large multi-page files, while maintaining performance and compatibility within the Spark ecosystem.