Founding Robot Learning Scientist - Embodied AI

Gram Games; San Francisco (CA); On-site; USD 225,000 - 300,000; Full time

Benefits offered by this job

Health coverage

Generous equity

On-site in Palo Alto

Relocation bonus

Job summary

GRAM is building reusable embodied intelligence and is seeking a leader for the robot-learning research program. You will set direction, evaluate architectures, and mentor the team to achieve transferable physical capabilities. The role focuses on data-grounded evaluation and scalable experimentation to guide next research directions.

Based in Palo Alto? No, the posting is for Palo Alto; this position anchors in a hands-on, on-site research environment.

Qualifications

• PhD in ML, robotics, CS, applied math, or equivalent with publications or deployable work.
• Led a consequential technical direction in robot learning or embodied models.
• Strong Python and PyTorch or JAX; working C++ for real-time inference.
• Trained an embodied model with a versioned dataset and evaluated on held-out configurations.
• Deployed a learned model on a physical robot or similar; documented results and failures.

Responsibilities

• Develop, pretrain, and adapt robot foundation models that acquire reusable physical capabilities.
• Set the research roadmap, standards, and experimental decision process for GRAM's robot-learning program.
• Design representations, objectives, architectures, and adaptation methods across morphologies, sensors, tasks, and environments.
• Build training curricula from heterogeneous physical and simulated experiences with strict dataset lineage.
• Scale experiments in PyTorch or JAX while isolating gains from architecture, data, compute, and initialization.
• Define evaluations and falsifiable capability claims for task transfer, robustness, and recovery.
• Deploy selected models through C++ robotics runtimes and analyze discrepancies to guide next questions.
• Help recruit, evaluate, and mentor researchers and engineers who extend the program.

Skills

Python

PyTorch

JAX

C++

Education

PhD in ML/robotics/CS or equivalent

Tools

Python

C++

Job description

GRAM is building reusable embodied intelligence and is seeking a leader for the robot-learning research program. You will set direction, evaluate architectures, and mentor the team to achieve transferable physical capabilities. The role focuses on data-grounded evaluation and scalable experimentation to guide next research directions.

Based in Palo Alto? No, the posting is for Palo Alto; this position anchors in a hands-on, on-site research environment.