Worked on the OpenHUTB/nn repository to enhance regression model performance, focusing on both linear and softmax regression pipelines. Applied data engineering and model optimization techniques using Python and TensorFlow to improve convergence speed, numerical stability, and overall accuracy. The technical approach involved refining convergence behavior and addressing numerical issues, resulting in faster training times and more reliable predictions across datasets. All contributions were linked to a single milestone and included co-authorship documentation for traceability. This work laid the foundation for broader production deployment and future optimizations, emphasizing robust engineering practices and collaborative development within the machine learning workflow.