I spent ten years three feet from the customer.Now I build for them.

I worked every position in a restaurant before I managed one: kitchen, bar, floor, and everything in between. A shift only works when people understand how their decisions land on everyone else.

On one Sunday opening, a grill had been left on high all night with no hood running. A coworker and I spent three and a half hours clearing smoke and wiping soot from every surface. Then I ran the floor until the evening manager arrived. What bothered me most was what a customer might walk into.

Managing meant P&L, scheduling, training, and getting more than twenty people through a Friday rush. It also meant jumping into someone else’s station, taking direction when they knew it better, and making a call when the room needed one.

The best shifts were never one person carrying the room. They were a group noticing what the others needed. I know how to lead that kind of team, and I know how to be part of it.

COVID ended that career in the spring of 2020, and I’ll be honest: I was ready for a change. The part I still miss is the crew.

What came next wasn’t software. I ran a t-shirt brand for a few years. It made sales when I fed it ad money and 80- or 90-hour weeks. The underlying math never improved, so I ended it.

I also spent a stretch of those years developing a protein bar with my brother. Hundreds of store-bought bars taste-tested, dozens of recipes. It never launched. I liked the customer and product work far more than the idea of running the operation.

I learned to ask a plain question: if I were seeing this work for the first time today, would the evidence justify another month? Sometimes the honest answer is no.

I grew up around computers and wanted to work in tech for a long time. AI finally gave me a practical way to learn by building products people could actually use.

My first setup was ChatGPT in one window and Xcode in the other: ask, test, read the error, try again. The tools have improved since then, and so has my understanding of the systems underneath them.

AI helps me write the code. I still have to understand the customer, define the product, choose the tradeoffs, test the edge cases, price it, ship it, and support it afterward.