I came up on job sites, not in a CS program. About seven months ago I wrote my first line of code. Since then I've shipped a contractor marketplace, trained a custom AI model for construction estimation, and built the tools I use to build everything else. People hear that and assume the jump from construction to software was huge. It wasn't. The two are closer than anyone gives them credit for.

A house and a codebase are the same problem

Framing a house is systems thinking. You sequence the work — you can't hang drywall before the rough-in passes inspection, you can't set trim before the floors go down. You estimate honestly, because a bad number doesn't just cost you margin, it costs you the next job. And you build to hold up under load, because a thing that looks finished but fails the first time it's stressed isn't finished.

Software is the same. Dependencies are sequencing. Architecture is load-bearing. A feature that demos well but falls over with real users is a wall that isn't tied into the frame. I didn't have to learn that mindset when I started coding — I'd been living it for years. I just had to learn the syntax.

What construction gave me that a bootcamp wouldn't

You finish what you start. On a site, an unfinished job is money sitting still and a client who can't move in. That instinct — close the loop, don't leave it framed and walk away — is rarer in software than it should be. Half-built features rot.

You respect the cost of every decision. When materials and labor come out of a fixed bid, you don't over-engineer. You build the thing that solves the problem and holds up, not the thing that's clever. That translates directly: the minimal solution that fully solves it beats the elegant abstraction nobody asked for.

You've actually lived the problem. This is the big one. My construction-tech isn't built from a product manager's user-interview notes. I've stood in the spot where a contractor loses a bid because his estimate took three days and the other guy's took an hour. That's why FairTradeWorker and ConstructionAI fit the trade — I'm not guessing at the pain.

How I actually learned

Not through tutorials. I learned by shipping, with real stakes. My first serious project had a real client — a contractor in Oxford, Mississippi — using it in production. Nothing teaches you faster than someone depending on the thing you built that morning.

I treated AI as a cofounder from day one. Not autocomplete — a partner I could think out loud with, that would catch what I missed and push back when I was wrong. That's a different relationship than most developers have with their tools, and it's the reason a guy off the job site could go from zero to a working AI estimation platform in months.

The transition wasn't a leap. It was the same work — build something real, make it hold up, finish it — in a new material.

I'm still early. Seven months is nothing. But I'd put the instincts I brought from the trade up against a CS degree any day, because the hard part of building software was never the syntax. It was knowing what "done" means and refusing to ship anything less.