Adelphi University
May ’22 – Dec ’24; Aug ’24 – Feb ’25
• Developed GNN models in PyTorch Geometric with 88% accuracy predicting semiconductor band gaps, supporting new materials discovery research.
• Built Polars-based ETL pipelines for 500K+ compounds, reducing processing runtime by 60% and enabling faster experimentation cycles.
• Optimized CUDA-enabled inference, improving model speed by 25% and lowering compute costs.
• Automated alumni survey reporting with Python ETL, reducing turnaround time by 75% and delivering Tableau dashboards that improved career-services outreach.