Senior Machine Learning Engineer, World Models

• Very strong coding skills in Python and PyTorch, or similar
• Deep experience training image or video generation models end to end
• A track record of running training at cluster scale
• Working knowledge of 3D vision, including multi-view geometry, scene reconstruction and physical priors

Our client builds task-specific world models and an evaluation platform for robot manipulation policies. They are a small, venture-backed team in San Francisco, founded by robot learning leaders with experience at some of the best-known names in AI, aerospace and autonomous systems.

Their world models simulate manipulation scenes accurately enough to validate robot policies, with the longer-term goal of training policies without needing a physical robot at all. You will develop the generative models behind this, with the controllability and physical fidelity needed to match how real robots behave.

Everyone on the team is an IC with deep ownership across a wide surface area. The culture is fast iteration and direct responsibility.

What you’ll do

• Train video and dynamics models, developing action-conditioned world models for manipulation policies
• Develop architectures and training methods that improve rollout quality over long horizons on hard physical tasks
• Own training infrastructure: run multi-GPU clusters, write custom CUDA and debug at scale
• Design, build and improve the data engine that lets the world model keep learning across customers and manipulation tasks