What it does
Medical supply runs on a hard constraint: a shortage is not an inconvenience, it is a patient without a drug. This system plans and reroutes supply to avoid that.
A hierarchy of agents handles planning and execution. Kafka moves the events, OR-Tools does the optimisation, XGBoost and Prophet handle forecasting, and a SimPy digital twin lets the whole thing be tested before it touches anything real.
Results
Measured against a heuristic baseline in simulation: logistics cost down 38.8%, and 98% of shortages prevented. Built to FDA 21 CFR Part 11 record requirements.
Note
This is my client work. I can walk through the problem little bit overview of system and code and the results.