Robotics Engineer: Foundation Model Training

1 мес. назад

СШАMiddleOnsite

pythonpytorchmachine learningreinforcement learningsimulation

Research Engineer role focusing on developing foundation models for robotic perception, action, and behaviors across various tasks.

We're hiring Research Engineers to develop foundation models for GRID. You'll work across model architecture, large-scale training, reinforcement learning, simulation, and evaluation to build robotic behaviors that generalize across tasks, environments, and embodiments.
We value exceptional depth in one or more relevant areas; candidates are not expected to have experience across all of them.
General Robotics is building the intelligence grid for physical AI — the platform that makes any robot, from robotic arms to humanoids, genuinely intelligent. Headquartered in Redmond, Washington, we're venture backed, including by Accenture, who invested in General Robotics in 2026 to advance Physical AI-powered robotics in manufacturing and logistics, and we're also part of Microsoft's Startups Pegasus Program. Our team's work spans some of the most widely adopted robotics and AI research to come out of Microsoft Research, Google Research and DeepMind — including AirSim, PACT, ClimaX, Tensorflow Object Detection and VideoPoet.
Our Singapore team works closely with the US team on GRID's research and development as an extended research facility, and to support our customers in the region. Engineers from both offices regularly visit each other for collaboration.

Formulate research hypotheses and design rigorous experiments informed by deployment needs.
Develop and scale foundation models for robotic perception and action.
Leverage simulation and diverse datasets for pretraining, post-training, reinforcement learning, and evaluation.
Scale existing simulation platforms for data generation and reinforcement learning across heterogeneous scenes and tasks.
Advance generalization beyond demonstrated tasks and robust spatial understanding.
Improve sim-to-real transfer and reliability under real-world deployment constraints.
Develop Auto-Engineering methods for scalable and robust deployment.
Implement, evaluate, and clearly communicate research as reusable, production-quality work.

Bachelor's degree in Computer Science, Robotics, a related technical field, or equivalent practical experience.
Research or applied experience in machine learning, robotics, or computer vision.
Strong Python programming skills and experience with PyTorch.
Experience developing and evaluating machine-learning models or robotics algorithms.
Experience with at least one robotics simulation platform.
Familiarity with inverse kinematics, dynamics, and robotic manipulation.
Demonstrated research ability through publications, substantial research contributions, or equivalent industry work.
Must obtain and maintain work authorization in the country of employment.
MS or PhD in Machine Learning, Robotics, Computer Science, or a related field, or equivalent industry research experience.
First-author publications at NeurIPS, ICML, ICLR, CVPR, CoRL, RSS, or ICRA.
Hands-on experience with Vision-Language-Action models, world models, world action models, diffusion or flow-matching policies, transformers, reinforcement learning, or imitation learning.
Experience with large-scale pretraining, data-mixture design, post-training, or reinforcement learning.
Experience scaling simulation platforms such as NVIDIA Isaac Sim or Isaac Lab, MuJoCo, or ManiSkill for learning and data generation.
Experience with sim-to-real transfer and evaluation on real robotic systems.
A track record of taking research from hypothesis through implementation and robust validation.