In March 2026, J. Park contributed to the apache/amoro and StarRocks/starrocks repositories, focusing on backend data engineering and system optimization using Java and Apache Flink. Park developed regex-based Prometheus metrics filtering to improve observability and reduce exporter noise, while maintaining backward compatibility. He enhanced AMS recovery logic to handle PENDING tables more robustly and fixed partition mapping issues following PartitionSpec evolution, ensuring correct data processing. Park also expanded Flink version compatibility and optimized StructProjection initialization for performance. In StarRocks, he improved Iceberg row size estimation with column-level statistics and refined incremental refresh logic to skip irrelevant snapshots.