About

I ran retail operations for twenty-five years before I built any of the technology those stores run on.

That order is the whole point. I have worked the task list a shelf system generates, and I have deployed that system into someone else's stores. Most retail AI is built by people who have never had to work the list.

Now I build the systems. At SymphonyAI that meant the analytics platform enterprise grocery and convenience pilots ran on, the lost-sales methodology that became the internal standard for what a shelf gap actually costs, an insight engine designed so the model never computes a dollar figure, and a test engine that audits what an AI app decides rather than what it renders. Where engineering owned the platform, I owned the methodology, the governance, and the semantic layer — the parts that decide whether a number survives being challenged.

Before that, Focal Systems, where I was close enough to the deep-learning team to run the labeling pipeline that trained the models and close enough to customers to ship Planogram Scan-in, the self-serve onboarding feature that fixed the cold-start problem every shelf-intelligence deployment hits. And underneath all of it, Wegmans: full P&L ownership and two decades of leading teams, which is where the operating discipline comes from.

I work AI-natively. I design, prompt, build, and ship with these tools every day; the work on this site is systems I run. I taught myself Python, TypeScript, React, SQL, and Docker by shipping them.

What I want next is the work that sits between the store and the system: hearing what is actually breaking on a shelf, turning it into a specification an engineering team can build from, and staying with it through deployment so that what ships is what was needed. Titles matter less to me than that translation.