AI Research Intern – Predictive World Model
Core
Develop predictive world models that learn physical world dynamics from multimodal data to serve as learned simulators for training and evaluating autonomous driving and robotic policies.
Role type
PhD research intern (generative modeling & world models)
Builds
Learned simulators for driving and robotic policy training/evaluation
Domain
Autonomous driving, robotics, generative AI
Deliverable
production ML models
Required skills
Generative modeling, Deep Learning, Computer Vision, PyTorch/JAX, Python programming, Large-scale data processing, Experimental methodology
Preferred skills
Video/3D generative models (diffusion, flow matching, NeRF, Gaussian Splatting), World models/learned simulators, Multimodal foundation models, Embodied AI research
Technologies
PyTorch, JAX, Gaussian Splatting, NeRF, Transformers, Diffusion models, Flow-matching models
Responsibilities
Drive end-to-end research projects on predictive world models; Design architectures for multi-view future prediction and generation; Extend predictions to 3D scene representations; Investigate cross-embodiment generalization; Build evaluation methodologies; Collaborate on scalable training pipelines and open-source results
Seniority
PhD candidate, Research Intern