About
Hi, I’m Jesús.
I’m an AI engineer and consultant. I spend the first part of every project making sure we’re solving the right problem, and the rest building the thing that solves it.
- Based in
- Bogotá, Colombia. US hours.
- Languages
- English and Spanish
- Experience
- 14 years in software, 10 in machine learning
- Author
- TensorFlow 2.0 Computer Vision Cookbook
The story
How I got here.
I started on the backend.
My first years were in insurance and fintech: secure storage, microservices, the first CI/CD pipeline one of those teams ever had. It taught me what production means. The system has to work on the day nobody is watching.
Then I taught machines to see.
I moved into computer vision and machine learning: people counters for retail stores, document AI for pharma, language models for legal text. In 2017 I started DataSmarts as a place to teach it, and in 2021 I wrote a book about it.
Then models learned to talk.
When large language models arrived, I went independent and started building with them: agents, retrieval, voice, and the evaluation that keeps them honest. Twelve production systems for ten clients since 2024.
Now I build the whole thing.
I sit with the people who have the problem, decide what is worth building, build it, and stay until it holds up. When a project needs a team, I bring one through DataSmarts, my studio.
How I work
Four things you can count on.
Problem first, model second.
Most AI projects fail before the first line of code, by solving the wrong problem. I spend the time to get that part right.
I’ll tell you when something is a bad idea.
Including when the cheaper option wins, or when you don’t need AI at all.
I design for when the model is wrong.
Evals, human review and guardrails come before the feature, not after the incident.
You’ll always know where things stand.
Short written updates, demos you can click, and every decision explained in plain language.
Outside work
Away from the keyboard.
I box, run and lift weights, and I’ll take a long drive over almost anything.