Hojin Lee

At Roadie, I build production forecasting and optimization systems that guide capacity planning and operational decisions across a national delivery network. My work spans demand, driver supply, capacity, delivery reliability, pricing, matching, routing, and batching.

I connect predictive models to real operating decisions, using temporal, geographic, retail-partner, vehicle, and delivery signals to forecast volume, fulfillment risk, and route feasibility. I’ve evaluated time-series methods, tree-based models, quantile prediction, and scenario analysis while focusing on stability through seasonal peaks and changing marketplace conditions.

I also strengthen the systems around models: Python and SQL pipelines, data validation, reproducible evaluation, historical backfills, monitoring, drift investigations, retraining, and deployment coordination across AWS, Azure, Docker, Kubernetes, and CI/CD.

Earlier, I applied quantitative analysis to lending at Happy Money, Global Lending Services, and TD, network optimization at UPS, and geospatial maximal-coverage modeling in research at Georgia Tech. I bring an industrial and systems engineering foundation to building decision systems that make complex operations more measurable and actionable.