Worked on expanding NonMaxSuppression compatibility in the llvm/torch-mlir repository to better support object-detection workflows. Developed support for center_point_box=1, enabling automatic conversion of bounding box formats from [x_center, y_center, width, height] to [x1, y1, x2, y2], which streamlines integration with upstream models. Added comprehensive end-to-end tests to validate the new behavior across the NMS path, improving reliability and reducing the need for manual format conversions. Utilized C++ and MLIR to implement these changes, applying machine learning domain knowledge to ensure robust feature integration. Focused on enhancing test coverage and maintaining code quality throughout the development process.