Hi! I'm Yohaï-Eliel Berreby, a Franco-Swiss researcher, engineer, and technophile, currently living in Montreal, Canada.

I'm trying to crack intelligence.

Research

I am currently a PhD student at McGill University's Department of Physiology and Mila, jointly supervised by Prof. Suresh Krishna and Prof. Audrey Durand. My research is currently focused on the intersection of AI neuroscience. How can the brain teach us to build better AI, and how can we use AI to understand the brain better?

My first peer-reviewed publication, "How forward remapping predicts perisaccadic biphasic mislocalization", was accepted at the Journal of Vision in March 2025 and is currently in press. For this project, I developed a high-performance JAX codebase to simulate systems of coupled DFT differential equations, with support for efficient interactive exploration of parameters. The code is publicly available here.

These days, I am working on bringing scalable deep reinforcement learning to computer vision.

A few things I've done

I've won hackathons (co)organized by Optiver, Meta AI, Entrepreneurs First, SCOR, Stability AI, l'X, Wavestone. I've also made podium in hackathons organized by Mistral AI, Orange Cyberdefense, and others. Some of these achievements were featured in official Télécom Paris press releases and highlighted by the then-Dean and President.

On the other side of the hackathon table, I was a Vice-President for the 2024 edition of PharmaHacks. I presided the Machine Learning committee, which designed neuro-AI and computational genomics challenges for 80+ students from across Canada, who were evaluated by a jury of McGill professors and myself.

I was the 6th employee of the Y Combinator startup Routine, which I joined in 2021 as an OCaml backend software engineer. I led the initiative that made it possible for Routine to use the same OCaml codebase on iOS as on other platforms, with fully-automated, type-safe, extensively-tested Swift <-> C <-> OCaml interoperability, and cross-compilation of OCaml code to aarch64-apple-ios (not officially supported at the time).

Before that, I worked on network and infrastructure engineering at C2S Bouygues. In the year I spent there, I automated the majority of the daily integrity checks in an IT infrastructure supporting the Bouygues Group's 80,000+ employees, designed, developed and deployed an internal data integration platform that cut down friction in inter-department operations within C2S, and worked on mission-critical infrastructure for the OnDijon smart city.

I hold a Diplôme d'Ingénieur from Télécom Paris, where I specialized in embedded systems (bare-metal and low-level software, FPGA/ASIC hardware programming with SystemVerilog).

More?

Want to chat? You can reach me via email, or send me a message on LinkedIn.

You can also find me on GitHub and Mastodon.