AI Research Engineer - Dexterous Manipulation (Egocentric Models)
Core
Lead the development of physical foundation models (Vision-Language-Action) using internet-scale egocentric video to enable humanoid robots to perform dexterous manipulation in unstructured environments.
Role type
Senior Research Engineer (Dexterous Manipulation & Foundation Models)
Builds
VLA foundation models, learning-based controllers, and generative action heads for humanoid robots
Domain
Robotics, Machine Learning, Egocentric Vision, Humanoid Systems
Deliverable
production ML models
Required skills
Egocentric vision, VLA modeling, Diffusion Models, Flow Matching, Reinforcement Learning, Robotics control, Python, PyTorch, Distributed training (FSDP, NCCL), Simulation environments (IsaacLab, MuJoCo)
Preferred skills
None explicitly stated
Technologies
PyTorch, FSDP, NCCL, IsaacLab, MuJoCo
Responsibilities
Architect and implement large-scale pre-training objectives for egocentric video datasets; Develop and scale multi-modal Foundation Models unifying visual perception and language; Design and optimize generative action heads using Diffusion Models and Flow-matching; Develop algorithms to align human-centric video representations with humanoid kinematic constraints; Utilize Offline RL and simulation fine-tuning to optimize model performance; Translate cutting-edge research into production-ready architectures
Seniority
Senior, hands-on IC
