About
A physicist who kept following the applied end of the problem. Here is the route, in order.
I am Miguel Torrez, and I build applied AI systems. I arrived at them through physics, and the order matters: the physics was not a detour taken before the interesting work started, it is the reason the work looks the way it does. Six years of it taught me to distrust a result I cannot reproduce, to strip a problem down until only its moving parts are left, and to keep asking what a number is actually measuring. I do exactly the same thing now with models and automation.
The oldest line on my CV is a school exchange I helped run out of San Salvador in 2015. A year later I was in Grenoble starting a physics degree, and I stayed for five more years: a bachelor, a magistère in mathematical physics, and a master's in astrophysics. In 2021 I spent five months at the Institute of Astrophysics and Planetology of Grenoble modelling the protoplanetary disk around TW Hydrae, and observed with the IRAM 30-metre radio telescope. I co-signed a paper. I also learned something less flattering and more useful: I enjoyed building the instrument more than defending the result.
So I moved sideways on purpose. A master's in international science trading taught me how deep tech is actually sold. A year at ALPAO, an adaptive-optics company, taught me what a technical datasheet is for and who reads it. A specialised master's in digital transformation at EM Lyon came next, then a year at Schneider Electric as an AI change manager, where the job was rarely to build a model and almost always to make several thousand people less afraid of one. X-HEC Entrepreneurs closed that loop in 2025.
Today I design and ship AI systems for a living: automation that runs unattended, retrieval and classification pipelines over messy multilingual data, and internal tools that quietly replace a recurring meeting. Most of that work belongs to the people who paid for it and will never be public. This site is the part that can be.
What lives here are the things I build to keep the physics muscle warm. A Lorenz attractor computed live in your browser, which you can heat up until it stops behaving. A three-dimensional graph of the entire French national curriculum: 6,051 micro-topics and 10,050 verified prerequisite links, from the first year of maternelle to the last year of lycée. Both come from the same instinct. Take a system that is normally described in paragraphs, and turn it into something you can pick up and rotate.
This site is published in six languages because the work crosses them. If you want to talk about a project, an experiment, or something someone has told you is impossible, write to me directly. I read everything, and I answer the messages that are actually about building something.
In short
- Now
- Applied AI systems, automation, and interactive simulations.
- Background
- Physics, mathematical physics and astrophysics at Grenoble Alpes University. Digital transformation at EM Lyon Business School. Entrepreneurship at X-HEC.
- Selected roles
- AI change manager at Schneider Electric. Marketing and communication at ALPAO. Research intern at the Institute of Astrophysics and Planetology of Grenoble.
- Builds with
- Next.js, Three.js and WebGL, Python, and large language model tooling.
- Published in
- English, French, Spanish, Brazilian Portuguese, Italian, German.
Get in touch
No form and no funnel. Email is the fastest route; LinkedIn works too.