Research Scientist - Robot Learning (VLA / WAM)

The role

You will own the training pipeline for vision-language-action (VLA) and world-action models (WAM) end to end. This role involves training policies that turn a world model into a robot that acts.

What you will do

• Own the training pipeline for VLA and WAM from data to a policy running on a robot
• Contribute to setting the technical direction for embodied research
• Close the sim-to-real gap through domain randomization, system identification, and calibration
• Build evaluation that predicts real-world transfer
• Adapt VLM backbones for control including encoder choice, adapter strategies, and co-training
• Curate and weight the training mix across heterogeneous robot datasets
• Design action representation and decoding including tokenization, chunking, diffusion, and flow-matching action experts
• Build world-model components that predict future observations conditioned on action
• Run post-training including supervised fine-tuning and RL for robustness

Required qualifications

• PhD in robotics, machine learning, or computer vision with a robot learning focus, from the PhD alone or followed by industry experience
• Publications at top venues such as CoRL, RSS, ICRA, IROS or CVPR, ICCV, ECCV, NeurIPS, open-source work, and/or deployed systems
• Deep experience with modern robot policy designs (VLA, WAM, diffusion) trained end to end
• Strong imitation learning fundamentals
• Familiarity with RL fine-tuning of pretrained policies
• Fluency with VLM backbones and how to adapt them for control
• Expert Python and PyTorch
• Multi-node distributed training experience (FSDP or equivalent)

About the company

SpAItial is pioneering the next generation of World Models, pushing the boundaries of generative AI, computer vision, and the simulation of reality. The mission is to redefine how industries generate and interact with physically-grounded 3D environments.