I am a research scientist at Facebook Artificial Intelligence Research, and a director of the Natural Intelligence ELLIS program , with Matthias Bethge and Peter Dayan .

My work currently mostly focuses on conversational AI and dialogue. Prior to that, I was a post-doctoral researcher and a Simons junior fellow. I worked with Nathaniel Daw, investigating computational models and neural bases of learning and decision making at the Center for Neural Science of New York University.

I obtained my PhD in 2012. I was lucky to have Yann LeCun and Jean Ponce as my advisors, and collaborate with other great people at those two places (mainly Francis Bach and Marc'Aurelio Ranzato). I was working on deep learning, machine learning, and computer vision as part of the Computational and Biological Learning Lab (part of the Courant Institute of Mathematical Sciences at New York University), and the Willow Team (part of the LIENS, Computer Science Department of Ecole Normale Supérieure, and INRIA). I also interned at Google Research Mountain View, under the thoughtful guidance of Samy Bengio. My first but deeply formative research experience was at the Gatsby computational neuroscience unit, working with Peter Dayan on trying to understand how serotonin and dopamine interact.

For up-to-date information: my Google scholar page.
Outdated CV here: (English or French).

Research interests

My main research interests are people-centered dialogue, learning, decision making, and machine learning. More precisely, my work has focused on:
  • Conversational AI.
  • Normative models of self-control, behavioral trade-offs between costs and rewards.
  • Deep learning and sparse coding in deep architectures for image recognition.
This webpage was shamelessly modeled after Armand Joulin's page, back when we were Willow students.