Contributed to the tenstorrent/tt-metal repository by developing comprehensive benchmarking documentation and data curation tools focused on GEMM and MatMull operations. Leveraging skills in data analysis, technical writing, and performance benchmarking, the work established reproducible workflows and clarified performance metrics across hardware configurations. Using Markdown for documentation, the developer detailed benchmarking methodologies, updated FLOPS data, and improved README clarity to support accurate and transparent performance evaluation. The approach emphasized reproducibility and traceability, enabling consistent measurement and informed hardware decisions. Over two months, the contributions laid a foundation for architecture-aware optimization and reliable benchmarking without direct involvement in bug fixing.