School of Engineering, Washington University in St. Louis | October 2025 – December 2025
In this course-based research project, I developed a Fairness-Weighted Maximum Weight Matching Algorithm for ride-hailing dispatch systems, motivated by the gap between Nash Social Welfare (NSW) guarantees in fair-division theory and the batch-based bipartite matching model used in practice (e.g., Uber). The algorithm embeds NSW-inspired fairness signals into classical maximum-weight matching by dynamically weighting drivers inversely to their accumulated utility. Furthermore, I established a batch-wise lower bound, proving that the FW-MWM algorithm consistently achieves a constant fraction of the optimal log-NSW improvement in each batch. I conducted extensive simulation studies across six market environments (heterogeneous utilities, probability imbalance, class imbalance, full and over-demand) and compared FW-MWM against Random and Match-and-Shift baselines. Results demonstrate near-perfect individual fairness, strong robustness under imbalance, and improved efficiency in heterogeneous markets.