Research Scientist, World Models

About the role

Waabi, a leader in Physical AI, is building autonomous trucks and robotaxis. As a Research Scientist in World Models you will develop and productionize next‑generation algorithms that reason about complex, dynamic 4D environments, focusing on large‑scale temporal reasoning, video, multimodal, and LLM/VLM/VLA models.

Key responsibilities

• Conduct fundamental and applied research in generative and predictive world‑modeling, including video generation, latent diffusion, autoregressive and flow‑matching models.
• Develop multimodal foundation models for driving scenes and LLM/VLM/VLA methods for scene understanding, reasoning, and control.
• Collaborate with engineers to integrate models into large‑scale distributed training and rendering pipelines.
• Publish high‑impact research at top conferences such as CVPR, NeurIPS, ICRA and mentor junior scientists.

Required profile

• Ph.D. in Computer Vision, Machine Learning, Robotics or a related field, or equivalent research experience.
• Proven ability to prototype and implement research using Python and PyTorch (or JAX) with strong software‑engineering fundamentals and experience in distributed training.
• Experience building generative or predictive models of the physical world for real‑world applications.
• Excellent communication skills and a track record of working in close‑knit research‑engineer teams.

Required skills

• Python programming.
• PyTorch or JAX deep‑learning frameworks.
• Distributed training infrastructure.
• Video generation and prediction techniques.
• Latent diffusion, autoregressive and flow‑matching models.
• Multimodal foundation models and LLM/VLM/VLA methods.
• Model distillation and compression.