AI Research Intern: Foundation Models

About the Role

GrayMatter Robotics is a well-capitalized Physical AI company headquartered in Los Angeles, building Factory SuperIntelligence (FSI) through systems that learn, adapt, and develop genuine process intelligence across every part they touch. Our robots take on the tedious, hazardous, and ergonomically demanding work that has kept manufacturing dependent on a shrinking skilled workforce, delivering superhuman performance. Robotics has a demo problem. GMR doesn't. Our systems are in active production today, adapting autonomously to real geometries, real materials, and real process variation without reprogramming. We deliver our system through a Robots-as-a-Service (RaaS) model that converts capital expenditure into an operating subscription, making Physical AI accessible to the manufacturers who need it most. We hire people who want to solve hard physical problems at an industrial scale.

What To Expect

• Work with the Special Projects AI team to work on next generation of foundation models that power GrayMatter Robotics Solutions (FSI)
• Take on real, complex, ambiguous problems and work through them on accelerated timelines, largely independently
• Develop foundation models for high precision manufacturing applications with multi-modal data beyond Vision or Force alone
• Validate models through rigorous testing on real world industrial robots and actual production parts
• Own the entire project lifecycle: data curation, simulation, model training, evaluation and robot deployment
• Write maintainable, production quality code for internal R&D

What You'll Bring

• Currently pursuing a PhD in Robotics, AI, Computer Science, Mechanical Engineering, Electrical Engineering or related fields
• Research experience in Foundation Models for robot manipulation applications
• Proficiency in Python and at least one deep learning framework: PyTorch is preferred
• Strong working knowledge and deep understanding of current World Models and Foundation Model Research
• Demonstrated experience working with simulation environments and/or physics-based modeling for robotics (e.g. Isaac Lab or MuJoCo)
• Experience with reinforcement learning, imitation learning, or policy optimization for robotic manipulation or process control
• Experience with robotics middleware and tooling (e.g., ROS/ROS2) and deployment on real robotic hardware
• A publication track record in venues such as ICRA, CoRL, RSS, IROS, RA-L, or T-RO, balanced with a strong bias toward real-world production impact

Nice to Have's

• Experience with VLA/VLM training
• Experience with industrial manipulators and demonstration of research deployed in industrial environments