Domdhi logoDOMDHI.OS
PUBLIC // ABOUT

About

Plan. Build. Verify.

I'm Dominic Baca, a senior software engineer with 15+ years of building software, based in New Mexico. I build AI agent tooling and the systems around it: I write the plan, AI agents build it, and tests and guardrails check the result. This page is how I think about the work, and what I've built so far.

Plan first

Some people dive into problems blind. I prefer reading the documentation before implementation. A good plan beats jumping straight in every time, and with agents it matters more, because they carry out exactly what the plan says. Plan for six days, and on the seventh you can rest while your agents take over.

Fix the cause

A closed ticket is not a solved problem. Close it fast, then find the root and make sure it can never happen again. Nearly six years of patching one application, one problem at a time, ended with a rebuild: 2,000+ obsolete files removed, every program a self-contained module, and tickets down from five to ten a day to a couple a month.

Software should work without its developer

Good software lets people finish their work without ever contacting the person who built it. Routine tasks like unlocking an account or removing an unapproved entry are self-service, and setup that took weeks takes a day. When something does break, the user reports it from inside the app and it reaches the right people as a ticket.

Ownership

Owning a system means treating it like my own. An error in it is my problem, and I take it personally. A codebase should hold only what has a purpose, because dead code makes a system harder to understand and never earns its keep. Code that might be needed someday isn't worth carrying, because if that day comes it can be rewritten with what we know now. The standard is what can be built today, not what was built years ago, no matter who built it.

Use every tool

Problems that are already solved should not be solved again. Hashing passwords is one: you use a proven algorithm instead of inventing a scheme, and you spend your effort on what nobody has solved yet. So look it up, ask the AI, and keep going until it works. Anyone who claims they don't is full of shit, and interview questions that pretend otherwise are a waste of time.

What I've built

An employee-services platform for a state agency, written by AI agents under my direction. Domdhi.OS, a personal operating system with a 264-tool MCP server over a 274-operation REST API. Workr, a live SaaS for solo tradespeople. Frontier Frenzy, a Western survivors game that runs in the browser on phone and desktop. And four public projects anyone can use: Domdhi.Agents, Domdhi.Doomstick, Domdhi.Crypto and Domdhi.Tach.

What I want next

Three things: software people actually use, a mentor whose experience I can learn from, and a team of high performers who share my passion for software and AI.