During March 2025, Thaink enhanced broadcasting robustness in the jax-ml/jax repository by developing a feature that enables reliable handling of mixed scalar and tensor inputs, including bfloat16, within JAX backends. Thaink improved type checking in the _bcast path and ensured scalar inputs are consistently converted to MLIR constants, addressing edge cases in numerical computing. In parallel, Thaink fixed a bug in ROCm/jax that corrected the conversion of Python native types and NumPy scalars to MLIR constants during MLIR conversion. This work, implemented in Python and leveraging compiler development expertise, deepened the reliability of broadcasting across diverse data types.