Melvin Cojon

About Melvin Cojon

I’m a junior at the University of Richmond majoring in Computer Science. Previously, I…

Experience

1. Machine Learning Researcher

   Capital One

   Oct 2023 — Present · Richmond, VA, US

   As a research assistant in the Parish Lab at the University of Richmond, I work on a publication-stage project using machine learning to predict lipophilicity, a critical property in drug design. My role centers on generating high-quality molecular features through advanced quantum chemistry simulations of molecules, then training and evaluating predictive models including XGBoost, LightGBM, Gradient Boosting, and Random Forest, achieving an R2 score of 0.80 with low prediction error.I built a custom Python pipeline to automate the preparation of molecular input files and extraction of properties from over simulation outputs, making the dataset clean, consistent, and ready for modeling. This automation has streamlined the entire workflow, from feature generation to model tuning, and has enabled targeted experimentation that improved performance across diverse molecule types.Beyond model development, I maintain consistency and organization across thousands of simulation files, contribute to literature review and figure creation for publication, and collaborate closely with researchers to troubleshoot computational issues. This experience has deepened my expertise in both applied machine learning and computational chemistry, while directly contributing to high-impact research outcomes.