Remote AI engineer for companies in Seattle, Washington

I build AI agents, the tests that guard them and the servers that connect them to company data, and I do it remotely from Monterrey, Mexico, for companies in Seattle. I am a Senior AI Engineer, I am not in Seattle, and my working day starts one or two hours before yours on Pacific time.

Seattle is on Pacific time and is known for cloud computing and software, for e-commerce, for aerospace, for biotechnology and global health, and for its port. I am in Monterrey, Mexico. This page takes those sectors one at a time and says what I would build for a company in each, which published work stands behind it, and where no case study on this site applies.

Pacific time, seen from Monterrey

Monterrey keeps the same hour all year. From early November to mid March Seattle is two hours behind Monterrey, and from mid March to early November it is one hour behind. Most of a working day overlaps in both seasons. The hours that do not overlap fall at the start of my day, before yours begins, so nothing you send in your morning waits for mine. What I produce sits in your repositories and your cloud accounts, where it is reviewed.

Cloud computing and software

A software company in Seattle does not need to be told what a language model is. What it tends to lack is time for the less visible parts: a set of cases that decides whether a new version of an agent ships, and tool access narrow enough to review line by line. Those are the subjects of LLM evaluation and MCP servers and integrations. The published work that transfers is open source, an Airtable MCP server that other developers use, and pre-registered research on agents.

E-commerce

Online retail runs on records kept in many places: catalog, orders, inventory, carriers, customer messages. AI helps when an agent can read those records together and act on one of them, for example by drafting the reply to a customer with the order in view. No case study on this site comes from e-commerce. What transfers is the work of bringing operational data into one place, written up as a data warehouse fed from operational systems. The service is AI agent development.

Aerospace, biotechnology and global health

For AI these sectors share one trait: their documents are controlled, and an answer must be traceable to its source. No case study on this site comes from aerospace or from biotechnology. What transfers is the method for answering from a controlled corpus, under RAG and data platforms, and the guide on agent auditability, which states what the record of an agent's action has to contain. For the plants and suppliers around aerospace, the closest published work is the platform of a lean manufacturing practice in Monterrey that works with food, metalworking and automotive plants.

The port

A port is logistics expressed in documents and messages: bookings, notices, exceptions. No case study on this site comes from a port either. What transfers is the same pair as in e-commerce: operational data joined in one place, and an agent that drafts the routine message for a person to approve.

If you write from Seattle

Use the contact page and describe the role or the problem. I answer in writing with what I would build first, the access it needs and what I would not promise. The form can be a contract under my name or a full-time senior role. What I cannot be is a person in your building. If the work depends on that, look for someone in Seattle.

Common questions

Do you live in Seattle?

No. I live in Monterrey, Mexico, and everything I do for a company in Seattle is done remotely. If the role needs a person on site every day, this page is not the answer to your search.

How far apart are Pacific time and Monterrey?

From early November to mid March Seattle is two hours behind Monterrey. From mid March to early November it is one hour behind, because Seattle changes its clocks and Monterrey keeps one hour all year. When your day starts at nine, it is eleven or ten here.

Our engineers already build with language models. What would you add?

Usually the parts around the model. That means a written description of the process, the tests that decide whether a version ships, and narrow, logged access to the systems the agent touches. A group that ships software every day often has the model working before it has any of those.

What would you build for an e-commerce company?

The useful first piece is usually an agent that reads the order and the customer message together and drafts the reply for a person to approve. Before that, the records have to be readable in one place. No case study on this site comes from e-commerce, so the page names the published work that transfers.

Do you have published work for aerospace or biotechnology?

No case study on this site comes from aerospace or from biotechnology. What transfers is the method: answers drawn from a controlled set of documents, a record of what an agent did, and tests that decide whether a version ships.

By city

Open to senior AI engineering roles and contract engagements, remote in the US, Mexico and Latin America.