Applied projects spanning forecasting, optimization, machine learning, and decision support.
Forecasting bank note demand across a central bank's regional distribution network to optimize inventory and avoid shortages or capacity overages.
Determining clinically and operationally optimal wait time targets using simulation, deep learning, regression, and inverse optimization.
Optimizing appointment scheduling for consult and follow-up visits using Markov Decision Processes and Approximate Dynamic Programming.
End-to-end regression pipeline comparing OLS, Random Forest, XGBoost, and LightGBM for real estate price prediction.
Multi-criteria decision analysis for pharmaceutical formulary inclusion/exclusion decisions using the UTADIS method.
Imbalanced classification pipeline for automating mortgage loan eligibility decisions based on applicant data.
Production recommender system for weekly offer personalization targeting loyalty customers, using transactional data and promotional calendars.
Production ML model to identify false in-stock signals in retail inventory, preventing approximately $13M in annual lost sales.
Decision-optimization system for staff scheduling and operational planning, delivering approximately $1M in annual savings.
Agentic retrieval-augmented generation system for product discovery and enriched customer interactions, using vector search and LLM orchestration.