Profile
About
I build Artificial Intelligence systems and test them until I trust them.
I’m an Assistant Professor at the Universidad Politécnica de Madrid, in the GAPS signal-processing group, where I also earned my Ph.D. cum laude in neural audio synthesis. My work turns deep learning into speech and audio models with knobs a human understands.
In 2025 I was a Postdoctoral Fellow at MIT’s Research Laboratory of Electronics. There I worked with Dr. Stefanie Shattuck-Hufnagel on acoustic landmarks in speech. In 2026, a Santander Postdoctoral Fellowship supports a new research stay at Queen Mary University of London.
Earlier, I co-founded a location-intelligence startup (now geogeogeogeo) and I build open-source tools like sota.ai, an LLM-assisted research assistant. Throughout, the same thing matters to me: that the model holds up outside the lab, and that someone can follow how it works. In a paper, in a lecture, or in a demo you can press play on.
Research interests
- neural audio synthesis
- voice cloning
- LLM tooling & agents
- geospatial AI
- model interpretability
- signal processing