Worked on the tensorflow/tensorflow repository to deliver module conversion passes that transform Shardy:MPMD modules into IFRT modules, targeting improved interoperability and performance of tensor operations across mesh configurations. The approach involved developing new utility functions and conversion patterns using C++ and MLIR, with Python used for supporting development and testing. A comprehensive test suite was implemented to validate correctness and maintain stability throughout the conversion process. This work established a foundation for future cross-IR compatibility and optimization efforts, enhancing maintainability and enabling more flexible tensor operation workflows within the codebase. No bug fixes were recorded during this period.