Full-stack engineer who builds AI into your product

I build the app end to end and the AI inside it: chat and voice agents, answers from your own documents (RAG) and LLM features, for U.S. and LATAM teams. One agent I built handles 1,200+ conversations a month.

Services for agents, RAG and full-stack apps

Three kinds of project, in Python and TypeScript, built on your own data and tools.

  • Chat and voice agents

    Agents that book appointments, answer questions and sell from your catalogue, on the web or the phone.

    LangChain, MCP
  • RAG and LLM features

    Answers grounded in your documents, with evals that show when a change makes them worse.

    Postgres, evals
  • Full-stack MVPs

    The web app, API and cloud setup the AI needs, built with Next.js, NestJS and AWS.

    Next.js, NestJS

See how a project runs

A voice and chat agent for a medical center

In production it handles 1,200+ conversations a month and has booked or moved 400+ appointments at 98% task accuracy.

Answers the clinic’s phone line

Speech-to-text, MCP tools and text-to-speech on a FreePBX line, so callers book or move a visit by talking.

Keeps the clinic’s calendar full

RAG over the clinic’s services in Postgres. The two agents cut operational work by 78%.

How a project runs

  1. A call to scope the work

    We go through the task, the data it needs and the systems it must reach, then agree on a first version.

  2. A prototype on your own data

    The agent runs on your real documents and tools early, so you judge real answers instead of slides.

  3. Evals, then production

    Each change runs against a test set of real conversations before it reaches your users.

9 years in production

Remote for U.S. and LATAM teams since 2017.

  • BairesDev

    Python analytics and a shared UI library

    2024–2026
  • Kualty

    Low-code builder, CRM backend and CLI

    2023
  • Treinta

    Led Next.js microfrontends and NestJS APIs

    2022–2023
  • Rootstack

    Real-time geolocation and AWS Lambda APIs

    2021–2022
  • IMUKO

    Live posture feedback from sensor data

    2021–2022
  • buenaPeople

    Tech lead for 5 developers

    2019–2020

Download my CV

Tools I ship with

Questions before you book

What usually comes up before a first call.

Book a call

What happens on the 15-minute call?

We go through the task, the data it needs and the systems it must reach. You leave knowing whether I am a fit and what a first version could look like.

How soon do we see it working?

The first milestone is a prototype on your real documents and tools, so you judge real answers instead of slides.

What if the agent gets things wrong?

Each change runs against a test set of real conversations before it reaches your users. The medical center’s agent runs at 98% task accuracy.

Can you work inside our codebase?

Yes. I write Python and TypeScript (Node.js, NestJS, Next.js, React) and join your repo, reviews and CI.

Can you help our team work with AI?

Yes. I have led teams of up to 5 engineers and bring Claude Code and GitHub Copilot into how teams build.