Model Research Scientist - Physical AI

What the role actually is

Kodiak is hiring a world model research scientist for physical AI in Mountain View. The role points at model-based understanding of driving scenes and autonomy behavior for a self-driving trucking platform.

World models matter because they can change how autonomous systems predict, plan, simulate, and evaluate what happens around them.

• Research world-model approaches for autonomous driving and physical AI
• Build experiments that connect perception, prediction, planning, and simulation
• Evaluate model behavior against real-world autonomy needs
• Collaborate with autonomy and product teams on how research informs the driving stack

• Research depth in machine learning, autonomy, robotics, or world models
• Experience with large-scale model training, simulation, or embodied AI evaluation
• Ability to turn research ideas into working autonomy experiments
• Strong communication across research and engineering teams

Kodiak has a focused autonomous trucking domain rather than a generic robotics platform. That makes the world-model work more grounded: success depends on whether the research improves a vehicle that has to operate in high-stakes logistics environments.