Your accounts, not mine
Anything that runs for your business uses the AI provider's API under your company's account. Never my personal subscription: provider terms don't allow it, and your data shouldn't go through it.
Remote · fixed price · USD
AI engineering for teams that need it to work, with prices on the page
I'm a senior engineer in Mexico City, working your hours. I rescue AI-built code that broke, put AI inside your product with permissions and logs, and train dev teams to work with coding agents without losing control. Every job has a fixed price agreed before work starts.
House rules
Model spend is separate and capped
What the AI provider charges is billed separately with a fixed monthly cap, measured in the first week of production.
A person approves the irreversible
Payments, refunds, emails to customers and deletions go through a human. Limits live in code, not in a prompt the model can ignore.
Every run is logged
The model reads or classifies once; the rest is deterministic code. Each step is recorded so you can see what happened and why.
Scope you can verify
Fixed price per deliverable, scope written as results you can check, and liability capped at the contract value.
How I work, stated plainly
I'm a senior engineer working with a team of supervised AI agents. That's why I can price below an agency; the review and the responsibility are mine.
Questions before we start
What does an AI code rescue include?
A review of the codebase and its deployment, a list of what's broken or unsafe ranked by risk, the fixes, and tests that prove each fix. Fixed price between $1,500–$3,000 USD depending on size, agreed after a free 20-minute scoping call.
Why fixed prices instead of hourly?
Because AI makes hours a poor measure of value, and you deserve to know the cost before work starts. Each price comes from my remote rate of $35–45 per hour applied to a scope we both sign off on.
Do you work US hours?
I'm in Mexico City, on Central Standard Time all year, so my workday overlaps most of a US one on either coast. Calls, async updates and handoffs happen in English.
What is an MCP server and why would I want one?
The Model Context Protocol is an open standard that lets assistants like ChatGPT or Claude query and act on your system with the permissions you define. Your team asks a question in plain language and the answer comes from your data, not from the internet.
Who owns the code?
You do, from the first commit: it lives in your repository and your accounts. If we stop working together, you keep everything.
How do you keep AI agents from shipping bad code?
Every change an agent makes is checked before it counts: tests that must fail before the fix and pass after it, independent review of claims, and nothing merged on the agent's word alone. Verification is the part of the work I don't delegate.
Tell me what's stuck
Two lines are enough: what you built, and what isn't working. I'll tell you which fixed-price option fits, or whether you need me at all.
07 / Contact
Tell me what's stuck and I'll scope it.
Fill out the form and I'll reply within 24 hours. You can also email me directly: