Building useful AI that speaks and listens.
Hi there! I am currently working at Kyutai (a non-profit AI research lab) as a Member of Technical Staff.
Before, I was a Research Scientist at Google DeepMind and worked as a Research Engineer at Facebook AI Research (FAIR).
My current research is focused on speech and speech-text LLMs. I co-first-authored the very first paper that applied Transformer-based generative language modeling on quantized speech representations. Fast-forward a few short years and this idea became one of the mainstream approaches for generative audio and ended up powering speech capabilities of the modern industrial-scale LLMs such as Gemini.
Other interesting speech-related projects I worked on:
Even earlier, I used to work on understanding the internal workings of deep learning models, studied communicating neural agents, dipped my toes in Federated Learning and a few other exciting topics!
I earned my PhD at the University of Glasgow.