Worked on the xupefei/spark repository to deliver a targeted memory optimization for Apache Spark’s query planning with wide schemas. Addressed SPARK-50229 by modifying the QueryPlan logic, replacing lazy val with def for AttributeReferences to reduce unnecessary object retention and lower memory pressure during the logical planning phase. This approach improved planning speed and reduced driver memory usage, directly enhancing scalability for large production workloads. The work was implemented using Scala and focused on backend development, demonstrating a deep understanding of Spark’s internals and memory management. No bugs were fixed during this period, with efforts concentrated on this feature enhancement.