Worked on the camel-ai/loong repository to enhance dataset and feature engineering pipelines, focusing on expanding physics data coverage and improving model reliability. Leveraged Python for data engineering and machine learning tasks, updating preprocessing workflows to deliver higher-quality inputs for downstream models. Addressed a bug in the physics verifier by strengthening symbolic multiplication handling and improving error reporting, which streamlined debugging and increased confidence in validation processes. The work emphasized robust dataset management and clear debugging information, resulting in reduced troubleshooting time and more accurate physics-based model validation. Demonstrated skills in software development, debugging, and scalable data pipeline design.