Research Scientist / Engineer - Data
Core
Develop methods to efficiently turn diverse data (on-robot teleoperation, simulation, off-robot) into better humanoid robot policies and shape data collection strategies.
Role type
Research Scientist / Engineer (Robot Learning & Data)
Builds
Humanoid robot learning systems and data processing workflows
Domain
Robotics, Machine Learning, Computer Vision
Deliverable
production ML models
Required skills
Data-centric research, robot learning, computer vision, multimodal learning, Python, PyTorch, experimental design, machine learning, data quality analysis, sequential data handling
Preferred skills
Egocentric data, robot teleoperation, 3D geometry, pose estimation, retargeting, cross-embodiment learning, imitation learning, reinforcement learning, sim-to-real transfer, active data selection, technical writing
Technologies
PyTorch, Python, simulation environments, camera systems, sensor streams
Responsibilities
Develop methods for off-robot data utility (action inference, retargeting, alignment), research data composition across training stages, design data curation tools, study source mixtures and supervision signals, run controlled ablations and scaling studies, measure impact on real-robot performance, shape collection priorities, work with data engineers on reproducible workflows
Seniority
Mid-to-Senior, hands-on IC
