During March 2025, Fum Chin refactored the operator test infrastructure for the pytorch/executorch repository, focusing on improving padding handling and test pipelines for critical operations such as AvgPool, Clamp, Clone, Conv1d, and Sub. Using Python and leveraging PyTorch’s backend development capabilities, Fum redesigned the test structure to enhance maintainability and align with ARM backend requirements. The updated pipelines reduced flaky test behavior and established a more robust foundation for future operator testing. This work improved test coverage, reliability, and feedback speed, supporting the project’s goals of dependable model serving and efficient developer workflows through careful pipeline design and unit testing.