Postdoctoral Researcher in Physics-Informed World Models and Agents

Postdoctoral Researcher in Physics-Informed World Models and Agents

UvA - University of Amsterdam

Amsterdam, Netherlands

€ 3.546 - € 5.538 per month gross

In this postdoc you will lead research on physics-informed (and physics-grounded) world models and agents: how to consolidate and transfer physical knowledge into learning systems for the real world. Physics engines and physics-informed neural networks are powerful, but the open problems are exactly where they stop being sufficient — under partial observation, in complex regimes that demand orders-of-magnitude speed-up, and where knowledge must transfer across embodiments, scenes and scientific domains. This is where world models earn their keep.

This is what you will do

You will help define and build the next generation of world models and the agents that act in them. Concretely, you will:

• develop physics-informed and physics-grounded world models that remain reliable under partial observation, where pure physics engines fall short;
• design methods that achieve large speed-ups in complex settings, so that agents can rehearse and plan at scale;
• study how to consolidate and transfer physical knowledge across embodiments, scenes and domains - from robot manipulation to biomedical and scientific dynamics;
• build and evaluate agents that act inside these models, with an emphasis on controllability, auditability and safety;
• lead and co-author publications at top-tier venues (CVPR, ICLR, ICML, NeurIPS, ECCV) and represent the lab at international meetings;
• help shape the research agenda, co-supervise PhD and MSc students, and contribute to proposals and collaborations (e.g. with industry and clinical partners).

What we ask of you

You are an independent researcher who can carry a line of work from idea to publication, and you enjoy mentoring others. You communicate clearly across disciplines, and you are energised - not deterred - by problems that sit between physics, learning and the messy real world.

Your experience and profile:

• a PhD (completed or near completion) in Machine Learning, Computer Vision, Robotics, Physics, Applied Mathematics, or a closely related field;
• a strong publication record at leading venues in machine learning, computer vision, robotics or scientific ML;
• demonstrated expertise in one or more of: physics-informed learning, world models, simulation/physics engines, dynamical systems, robot learning, or scientific machine learning;
• excellent programming skills (e.g. Python, PyTorch) and experience running substantial experiments;
• the ability to drive research independently and to mentor PhD and MSc students;
• a professional command of English; willingness to learn some Dutch is welcome but not required.

It is a plus if you additionally have hands-on robotics experience, experience with differentiable/GPU-parallel simulation (e.g. Isaac Lab, MuJoCo MJX, Genesis), or experience applying physics-informed methods to biomedical or scientific data.