Contributed to the pytorch/executorch repository by developing a decomposition pass for the fmod operator on the Arm backend, targeting optimization of floating-point modulus operations. The implementation focused on enhancing numeric correctness and backend robustness, laying the groundwork for future Arm-specific performance improvements. Comprehensive tests were created to cover a wide range of edge cases, including special values such as NaNs, ensuring reliable behavior across diverse inputs. This work leveraged Python, PyTorch, and algorithm design skills, with an emphasis on backend development and thorough testing practices to improve the efficiency and correctness of floating-point operations on Arm architectures.