Worked on the pyg-team/pytorch_geometric repository to enhance data quality and dataset integrity, focusing on the MD17 dataset. Addressed a critical bug by correcting the spelling of 'Naphthalene' in both documentation and internal dictionary keys, ensuring that dataset labels accurately represent the intended chemical compound. This fix improved the reliability of data curation and dataset management processes, reducing the risk of labeling errors during data loading and model training. Utilized Python to implement changes that consolidated documentation standards and maintained robust traceability, supporting future audits and minimizing the likelihood of similar issues in subsequent releases or experiments.