Large-Scale Exam Scheduling via Integer Programming
Applied Optimization Project, Cornell University, 2023
- Leader of the Prelim Scheduling subteam for Cornell Scheduling Team using integer programming.
- Incorporated fairness constraints and instructor preferences into the optimization model.
- Scaled to 10,000+ decision variables using real-time Gurobi checkpointing.
- Decreased two-exam student conflicts by 16.5% through constraint refinement.
- Built a 6-layer MLP (≈3k parameters, 95% accuracy) to predict post-add enrollment.
- Collaborated with university administrators to iteratively refine constraints and objectives.