Developed the initial scaffold for an autonomous driving semantic segmentation project in the OpenHUTB/nn repository, focusing on reproducibility and streamlined onboarding. Leveraged computer vision and machine learning techniques to establish a U-Net baseline for addressing class imbalance in semantic segmentation tasks, utilizing CARLA for simulation data. The work included creating a comprehensive README in Markdown, standardizing naming conventions, and restructuring the project folder to autodriveseg for clarity and consistency. Emphasis was placed on clear documentation and accessible project structure, enabling faster experimentation and collaboration. No bug fixes were recorded during this period, with efforts concentrated on foundational feature development.