Paul Pacaud

I am a PhD student at INRIA Paris, ENS PSL working at the intersection of Computer Vision, NLP, and Robotics. My recent work focuses on Multimodal Reasoning, Synthetic Data Generation, and Vision Language Action Models. I graduated from CentraleSupélec, and did robotics research at Stanford. My long-term aspiration aims to push the boundaries of AI’s ability to perceive, understand, and act in the physical world.

I’ve worked in very diverse environments ranging from applied science in industry (quant trading @ Engie, cloud consulting @ PwC), product development in startup (startup @ Brio) and academic research (PhD studies @ Inria). Here are a few takeaways:

At Brio, the startup I co-founded, I built and deployed an AI-powered interview coaching platform for universities. Brio helps students get an interview (CV analysis, Linkedin analysis, Cold reach automation), and then helps you prep for this interview (interview simulation). Brio is now helping +15,000 students across +20 paying partner universities (among them HEC, Science Po, EDHEC, CentraleSupélec…).

At Engie Trading, I worked within the Power & Gas Trading desk. I helped to approximate hard constrained optimization problems on the Power Market, and deployed applied ML models on the Gas Market.

At Inria and École Normale Supérieure Paris, I mostly work on VLMs for Robotic Manipulation, around three subtopics: Multimodal Reasoning, Synthetic Data Generation, and Vision Language Action Models.

Please feel free to reach out to me if you want to chat!