PhD in Reactive World Models for Autonomous Driving École Polytechnique in France

École Polytechnique and Stellantis are advertising a 3-year CIFRE Industrial PhD in Reactive World Models for Autonomous Driving, based in Paris, France. The project sits at the intersection of streaming video generation, world models, autonomous driving simulation, and computer vision, with a strong focus on building controllable, photorealistic long-horizon video models that can respond to driving actions in real time.

The PhD is jointly hosted by Stellantis (with Hai Li) and the VISTA team at École Polytechnique. The role is described as a CIFRE PhD, meaning the student will be employed by Stellantis while carrying out research in collaboration with the university. The post highlights access to a large-scale internal GPU cluster, real-world driving data, and industrial simulation pipelines.

Research areas and keywords: autonomous driving, world models, video generation, deep generative models, simulation, computer vision, robotics, applied mathematics, and AI. The work is especially relevant for candidates interested in learning-based perception and prediction, generative modeling for dynamic environments, and control-aware video synthesis.

Eligibility highlights: the ideal candidate has a Master's degree in CS, CV, AI, applied math, or robotics, with a strong background in computer vision and deep generative models. Prior experience with generative models, video generation, world models, or simulation is considered a plus. Applicants should also know Python and PyTorch, and be fluent in English.

Application window: the post states a September 2026 start date, but no deadline is given in the post. Interested candidates should consult the official job description for the full application instructions.

How to apply: open the official job description linked in the post, review the detailed requirements, and follow the application instructions there. The direct application link shared is included below.