During November 2024, Siying Dong focused on enhancing the xupefei/spark repository by addressing a stability and performance issue in Spark Streaming. Siying optimized the stream-stream join logic to conditionally fetch checkpoint information only when the operation is supported, thereby reducing unnecessary processing and preventing assertion failures in edge cases. This work, aligned with SPARK-50253, improved runtime efficiency for streaming workloads by limiting checkpoint fetch operations to relevant scenarios. The solution was implemented using Scala and leveraged expertise in Apache Spark and stream processing, demonstrating a targeted and thoughtful approach to resolving a nuanced bug in distributed streaming systems.