Worked on the apache/beam repository to implement dynamic runtime selection of the streaming engine harness, enabling the system to adapt automatically to varying job connectivity types. Developed a configuration-driven switching mechanism that selects between FanOutStreamingEngineWorkerHarness for direct-path workloads and SingleSourceWorkerHarness for cloud-path scenarios, all without requiring redeployment. This approach leveraged Java and core distributed systems concepts, utilizing Beam and Dataflow to enhance streaming reliability. The feature reduced manual intervention and improved throughput and latency stability by allowing seamless adaptation to changing network conditions, addressing both resilience and operational efficiency in streaming systems with a single, traceable code change.