Worked on the pytorch/executorch repository to deliver a CoreML backend placeholder aimed at stabilizing ongoing development. Addressed a persistent issue where uquery-related failures disrupted CI and test cycles, implementing a Python-based backend scaffold that prevented these failures and maintained momentum for downstream improvements. The approach focused on backend development principles, ensuring that the placeholder would not block future integration of the full CoreML backend. By establishing this interim solution, the work enabled the team to continue iterating on other features without interruption, while providing a clear technical path for eventual CoreML backend support within the project’s architecture.