Hey there. My name is Fabian Pedregosa and I'm a senior research scientist at Google. This is convenient because it allows me to do a living from two of the things that I love the most: math and computer programming.
I work on algorithms to make machines learn better. How did I get into this rabbit hole? Glad you asked.
Presenting scikit-learn at FOSDEM 2011.
Originally from Madrid (Spain), an expired Mathematica license during my college years at the University of Granada (Spain) pushed me to become a core contributor to SymPy (computer algebra system in Python). I immediately got hooked by the power of open source collaboration.
I couldn't believe, and it still strikes me today, that people would be giving away high quality software, and taking time to fix bugs and mentor potential contributors for a greater good. Open source is nuts.
Anyway, after graduation I moved to Paris (France) to work at INRIA on another open source project: scikit-learn. I had contributed to open source before, I could't believe I could make a living out of that!.
A couple years later, I felt the best way to continue contributing to open source was through academia. I started a PhD that took me to applications of machine learning to brain imaging. This was much harder than I expected but I eventually graduated.
Orientation in Google Montain View, October 2018.
After that I drifted more towards mathematical optimization, which is my current main research topic, with a Postdoc at INRIA Paris (with Alexandre D'Aspremont) and then UC Berkeley (with Laurent El Ghaoui).
Visiting a cheese manufacturer in Asturias, Spain, in the summer of 2022.
After my period at Berkeley I joined Google Research in Montreal. There I have the chance to work with incredibly talented colleagues and students.
See my CV for more details (last updated Dec. 2021)
When I'm not working, I enjoy cycling, playing guitar and cooking with my wife. I currently live in the beautiful city of Geneva, in Switzerland.
Email (f@bianp.net) is the best way to reach me. You can also follow me on twitter and Mastodon.
I enjoy mentoring young students and chatting with colleagues, so if you have an idea that you'd like to discuss with me, don't hesitate to shoot me an email.
Most of my code can be found in Github. You'll find all of my papers on ArXiv and Google Scholar.
I'm currently not taking interns or students. I will update this page when I'm available again.
I enjoy contributing code to open source projects. Here are the ones I am most proud of:
Optax is the most used optimization library for JAX.
Jaxopt is a hardware accelerated, batchable and differentiable optimization library based on JAX. In December 2023 this project was discontinued in favor of Optax
David Cournapeau started this project in 2007. I took over the project in January since 2010, made the first public release the 1st February 2010 and have been developer, maintainer and release manager until January 2012. I currently contribute ocasionally.
line-by-line memory consumption for Python code. I started this project in 2012 and are the current maintainer and lead developer. This project won the best poster award at the EuroScipy 2012 conference.
SymPy is a Computer Algebra System written in Python. Originally written by Ondrej Certik, I was the first contributor to join the project. I’ve made important contributions to this project in the period 2007-2009, including a Google Summer of code scolarship in 2010 to improve the logic module.
I have contributed some code to improve the linear algebra module, including the
linalg.solve_triang
function introduced in 0.9 and some improvements to orthogonal
decomposition appeared in the 0.10 series.
I support the Gnome desktop financially through the friends of GNOME program.
I am married with Valentina Borghesani. She is a great inspiration on so many levels, and I owe her so much.
Until 2010 I played in a pop band called Los Esclavos. Available records are
And before that, I had a band called DoDo, and made a record