About

I'm Clément — part engineer, part entrepreneur, part (mad) scientist, building AI at Google DeepMind. In the past 20y, I've been driven by an obsession: figuring out how to build AI systems that can learn on their own and ultimately redefine how we write software.

My research started in artificial vision in the early days of deep learning at NYU with Yann LeCun: from designing e2e trainable vision systems via conv neural networks and custom low-power dataflow computers (NeuFlow), to co-architecting Torch (PyTorch's ancestor) with Ronan Collobert and Koray Kavukcuoglu — and then bringing deep learning into production through Madbits, Twitter Cortex, and NVIDIA (MagLev, RAPIDS).

Today at Google DeepMind, I lead the AI Platform & Models teams — scaling the serving infrastructure, developer APIs (Interactions API), and agentic harnesses (Antigravity) behind Gemini, alongside our open model family, Gemma. As these systems cross the threshold from imitation to invention, my focus is on turning frontier models into reliable, general-purpose agents that can reason, collaborate, and act across tools — building at full pace, with full accountability.