Worked on the numpy/numpy repository to address a critical bug in the np.unique function, specifically improving how NaN values are handled when the equal_nan parameter is set to True for one-dimensional arrays. Applied Python debugging and data analysis skills to ensure that NaN values are now correctly collapsed, aligning the function’s behavior with user expectations and enhancing the reliability of data cleaning workflows. Added comprehensive regression tests to validate the new behavior and performed test suite linting to improve maintainability. This work reduced support risk and confusion for users relying on np.unique in their data processing pipelines.