Baptiste Méjean
Computer science student at EPFL and the University of Washington, researching world models and latent-space planning for robotics.
Background
I want to understand how machines can understand the physical world, reason about it, and act within it.
What I really want is for machines to reason better. Today's models are impressive, but they still struggle with cause and effect, and with planning more than a few steps ahead. Getting that right is what will let AI make a real difference in places like hospitals, factories, schools and in systems people rely on to stay safe. That's the problem I want to spend my time on.
I got into technology because of what it has done for people and what it can still do for them, and that hasn’t changed. The projects I care about most are the ones that free people from dangerous and repetitive labor, improve their lives, and give them more time for what they truly care about. It's how I decide what to work on, and it's what keeps me going when things get hard.
Right now I'm finishing my Computer Science degree at EPFL, and spending my final year at the University of Washington, where I do research in the RAIVN Lab. When I'm not working, I'm usually making music, playing basketball, or looking at art.
Research
Interests
Experience
Oct. 2026 – Present
Undergraduate Researcher
RAIVN Lab, University of Washington · Seattle
I'm leading an independent project on latent-space planning for robotic manipulation. The idea is to develop planning for world models with object-centric representations, so a robot can reason about the individual objects in a scene and plan its actions in a learned space rather than in raw pixels.
In practice, that means extending the lab's 3D hand-object forecasting pipeline to goal-conditioned planning for a gripper robot, trained on motion-capture data with RoboCasa and DROID. I'm comparing action-conditioned two-pass planning against an action-only single-pass approach, with the goal of taking whichever works best to dexterous hands. Mentored by Rustin Soraki.
Jun. 2026 – Sep. 2026
Engineering Assistant
Advanced Machine Intelligence (AMI Labs) · Paris
I built the demo app for one of the company’s first research-to-product prototypes and helped develop the streaming infrastructure powering all of its live applications, running inference at 20 fps with closed-loop evaluation on live video. I worked closely with AMI’s researchers and contributed to the research decisions behind the prototype.
Education
Sep. 2026 – Jun. 2027
Exchange, Senior Year
University of Washington · Seattle
Paul G. Allen School of Computer Science and Engineering
Coursework in Machine Learning, Deep Learning and Computer Vision
Research @ RAIVN Lab
Sep. 2024 – Jul. 2027
BSc, Communication Systems and Computer Science
EPFL · Lausanne
IC Faculty
Coursework in Machine Learning, Discrete Mathematics, Linear Algebra, Probability & Statistics, Programming, Algorithms and Signal Processing
Work
Apr. 2024 – Present
Part-time Full-Stack Web Developer
Action Marketing Services · Paris
I built a notification service for an event planning company.
Sep. 2025 – Feb. 2026
Teaching Assistant, Advanced Information, Computation, Communication
EPFL · Lausanne
I led exercise sessions and quizzes for a class of 800 students.
Jan. 2023 – Jan. 2026
Founder
Goodminds · Paris
I built multimodal deep learning systems for automated social media moderation and misinformation detection.
Jun. 2022 – Jul. 2022
Intern
Nabla Technologies · Paris
I built NLP pipelines that map free-text medical data onto structured medical taxonomies.
Projects
