AI engineer

This is not a single opening. When a US company asks for this skill, we introduce engineers from the network. Active client openings are on LinkedIn.

What this work is

You put a model feature inside a product a US team already ships. The work is the prompt or the tool call, the product code around it, and a way to tell when the output is wrong. A demo notebook is not the feature.

The stack

• The language the product is written in
• A model API
• Prompts and tool calls
• A review set or an eval for bad outputs
• Logging when the model is wrong
• The product path the feature has to live in

What we look for

• A shipped feature that calls a model, not only a notebook
• Prompts, tools, and a way to tell when the output is wrong
• The product stack the feature has to live in
• 5 or more years as a software engineer

How we check

We review work you have shipped. A keyword on a resume is not enough. You need advanced or fluent English. You will be in standups, code reviews, and Slack with a US team. You are assessed at the top of the four-level scale: you design the model workflow and you can explain a bad output. Every engineer is assessed on the four-level AI fluency scale before a client meets them. See the four levels.

How a placement runs

You join the client team and work in their repo, their standups, and their code reviews. You work US business hours from Mexico, Costa Rica, Colombia, Peru, Brazil, Chile, Argentina, or Uruguay. The placement is almost always a full-time contract. You are paid the same amount every month in US dollars. The monthly invoice is for bookkeeping. Candidates never pay to apply or to join. How joining works.

Hiring AI engineers.

Updated November 9, 2026.