Worked on enhancing the NVIDIA/NeMo-speech-data-processor by improving the documentation of audio quality assessment metrics within the codebase. Focused on metrics.py, the work involved providing detailed explanations for PESQ, STOI, and SI-SDR, clarifying what each metric measures and offering guidance for interpretation. This documentation-first approach, using Python and technical writing skills, aimed to reduce ambiguity in data evaluation and streamline onboarding for new engineers. By aligning metric definitions with the processing pipeline, the contribution improved maintainability and reproducibility for future development. No bug fixes were required, as efforts centered on strengthening clarity and supporting data quality workflows.