Soowon worked on improving the HLO parser within the Intel-tensorflow/tensorflow and Intel-tensorflow/xla repositories, focusing on correcting the window attribute dilation field name from rls_dilate to rhs_dilate. Using C++ and leveraging strong bug-fixing skills, Soowon implemented a unified solution that addressed parsing errors and prevented misleading error messages in the HLO pipeline. The changes ensured consistent behavior across both codebases by applying targeted commits, which improved the reliability of window attribute parsing. This work contributed to more robust downstream optimization and JIT compilation processes, reducing developer time spent on triage related to window attribute parsing issues.