Robot Foundation Model Engineer

Responsibilities

• Design Mobile Manipulation VLAs that integrate navigation and implement control, and define safety and evaluation criteria
• Build training pipelines combining pre-training, imitation learning, RL post-training (including RLVR), and Sim2Real
• Build simulation-based closed-loop and dataset-based open-loop evaluation systems, and validate models in the field
• Analyze field logs and failure cases to develop roadmaps for model architecture and training data improvements
• Collaborate with control, simulation, data, and PM teams to bring models into service at customer sites

Qualifications

• Broad understanding of AI/ML with expertise in at least one area
• Implementation skills in Python / C++ and hands-on system development and experimentation experience
• Experience taking a system through experimentation, deployment, and improvement
• Experience incorporating inference latency, model size, memory, and safety constraints into model and training design
• Ability to systematically analyze and define failure cases from the field

Preferred Qualifications

• Experience training LLMs or VLAs, or using RL-based post-training methods such as RLVR
• Experience deploying models on real equipment, vehicles, or robots, including brownfield environments, noisy sensors, and edge cases
• Experience integrating motion planning, SLAM, localization, and control
• Experience with ROS2, edge inference, and on-device optimization
• Experience with safety validation, fallback logic, and evaluation harnesses
• Publications at major ML/robotics conferences or journals, or awards in science and engineering olympiads

Who We Are Looking For

• You start by asking which technology best fits the problem
• You want to close the gap between benchmark accuracy and field performance
• You design for both current customer value and future scalability
• You want to build reusable product capabilities rather than one-off solutions