Research Associate - World Models & Video AI for Robotics

HireArt, Inc.; Los Altos (CA); On-site; USD 60,000 - 80,000; Part time

Job summary

Toyota Research Institute (TRI) is hiring a Research Associate to join the Learning from Videos (LFV) team. You will advance world models, video generation, and policy learning, contributing to a shared research codebase and publishing in top venues.

The role emphasizes building state-of-the-art baselines, training large-scale models, and evaluating performance in simulation and on real robotic hardware. The ideal candidate is a current PhD student with strong foundations in computer vision,

Qualifications

• Current PhD student in a related field.
• Strong fundamentals in computer vision, video understanding, generative models, 3D reconstruction, or robotics.
• Experience with video diffusion models, world models, and world-action models.
• Proactive and self-directed with the ability to work in a research-driven environment.
• Strong communication and collaboration skills and ownership of problems end-to-end.

Responsibilities

• Contribute to research on multimodal and multi-view world models, video generation, video policies, and related areas; coordinate with university partnerships, research meetings, and publications.
• Maintain and evolve the Any4Dv3 video generation codebase, review PRs, stress-test capabilities, and develop new functionality.
• Benchmark state-of-the-art methods for world-action models (WAMs) within Any4Dv3; deploy in simulation and on real hardware; identify gaps and develop solutions.
• Produce maintainable, well-documented code and contribute to internal tooling and open-source releases.

Skills

Computer vision

Video understanding

Generative models

3D reconstruction

Robotics

Education

PhD student in related field

Tools

Any4Dv3

Job description

Toyota Research Institute (TRI) is hiring a Research Associate to join the Learning from Videos (LFV) team. You will advance world models, video generation, and policy learning, contributing to a shared research codebase and publishing in top venues.

The role emphasizes building state-of-the-art baselines, training large-scale models, and evaluating performance in simulation and on real robotic hardware. The ideal candidate is a current PhD student with strong foundations in computer vision,