Worked on the WebKit/WebKit repository to improve audio capture reliability by addressing a critical issue in device enumeration. Focused on refining the process to exclude speaker devices from capture sources and ensure unique device identifiers by incorporating the GStreamer class, which prevented misclassification and broken GStreamer pipelines. This solution reduced exclusive access conflicts and enhanced the stability of audio pipelines across different environments. The work involved C++ development and leveraged expertise in audio processing, device management, and the MediaStream API. The approach demonstrated attention to cross-platform compatibility and careful handling of hardware resources, resulting in a more robust audio capture workflow.