Live 3PL deployments and seven-figure outcomes. Currently at Americold; building toward the intersection of applied OR and AI in supply chain operations.
Most operations problems are stated wrong on first reading. The leverage usually lives in a framing nobody has questioned — a column in a report meaning something different from its name, a daily average blending three populations, a "demand" curve that's actually a processing rate.
Formulation, validation, and deployment cycles compressed from quarters to weeks. The mathematical rigor stays intact; the syntax gets generated under direction. This is what AI-accelerated means in practice — not a vendor platform claim, a workflow.
The warehouse floor, where standard work has to survive contact with a real operator. The GM's office, where the analysis has to translate into a decision. Peer review, where the methodology has to defend itself against people whose job is finding the flaw. The work has to land in all three.