Worked on the numpy/numpy repository to enhance the POWER VSX feature mapping within the NumPy testing framework. Developed logic to dynamically assign power-generation flags based on detected platform capabilities, refining architecture-aware feature mapping and improving the accuracy of test detection. This approach reduced false positives and negatives in POWER-based continuous integration runs, leading to more stable and reliable test outcomes. The work involved Python, regular expressions, and subprocess management, with a focus on dynamic flag logic and robust testing practices. These improvements contributed to smoother release cycles by enabling faster, platform-agnostic validation and reducing test flakiness across diverse environments.