One wrong join changes a regulator's read of the numbers.

That's the kind of work I did for ten years at a major financial institution: federal regulatory reporting, attribute-level data lineage, migrations where the number on the last page has to survive an audit. It teaches you a specific habit — you don't trust an answer until you can trace it back to source.

I brought that habit to AI. On the side of that job I built an internal AI tool that grew to 140+ users through word of mouth, with no mandate and no budget, and turned multi-day workflows into minutes. Since then I've built five live systems on the same spine: facts in code, judgment in the model, and a verifier in between so nothing reaches a person unchecked.

I'm Matt Price. CIS degree from Tarleton State, a decade of construction before any of it. Everything here is built solo, end to end — data model to front end.