Worked on improving the reliability of core linear algebra operations in the numpy/numpy repository by addressing in-place aliasing issues in matrix-vector and vector-matrix multiplication routines. Used C and Python to implement overlap-safe handling, ensuring that when input and output arrays overlap, results remain correct and silent data corruption is avoided. Added comprehensive regression tests to verify the robustness of these changes and aligned the behavior of matvec and vecmat with matmul’s established overlap semantics. This work enhanced the correctness and stability of NumPy’s numerical computing routines, supporting more dependable scientific workflows and increasing developer confidence in downstream applications.