Research Scientist, Robotics

Description

The company builds robots that can handle physical work, aiming to bring robots into factories, warehouses, homes and businesses, where few exist today. Nearly 9 in 10 American factories and warehouses have no robot at all.

The company is a robotics research and deployment business. It builds general-purpose robot hardware and AI, including a semi-humanoid robot with near-human dexterity and a robot that moves boxes, pallets and bulk goods for customers ranging from family-owned manufacturers to some of the world's largest brands. Robots are built at scale and learn as a connected network, so each robot's experience improves the whole fleet's AI.

Founded at MIT and based in Watertown, MA, the team designs and makes its own hardware, trains its own models, works directly with customers of every size, and maintains its robots across the United States. The team of 160+ includes researchers, engineers, technicians, operators and salespeople.

Responsibilities

• Push the limits of what robots can do across the whole stack, from hardware and data to models and deployment.
• Use a fleet of 100+ robots, a large pool of teleoperators, in-house manufacturing, and the ability to gather real-world data at very large scale for data collection and evaluation.
• Set the research direction for how the robots learn, covering the full path from behavior cloning to supervised fine-tuning and reinforcement learning.
• Frame dexterous manipulation problems and produce results that hold up when run on real robots.
• Develop methods that learn from large volumes of human input, including teleoperated demonstrations, interventions, corrections and human feedback used as reward.
• Design experiments across the robot fleet and build evaluations that tell real gains apart from noise.
• Train and post-train vision-language-action models using behavior cloning, supervised fine-tuning and reinforcement learning.
• Partner with research and applied engineers to turn new techniques into models that run on the robots already deployed with customers.
• Document findings in internal write-ups and technical reports.

Requirements

• At least 5+ years of research experience in robot learning, machine learning or a related field, gained through a PhD or equivalent industry work.
• A record of original work in robot learning, imitation learning, reinforcement learning or large-model training, shown through publications, open-source projects or shipped systems.
• Strong Python skills and hands-on experience with PyTorch or JAX.
• Experience running experiments on physical robots, not just in simulation.
• Ability to communicate results clearly, including negative ones.
• Nice to haves
• Experience with vision-language-action models, or with diffusion and flow-matching policies.
• Experience learning from human feedback, interventions or reward modeling.
• Experience with bimanual or dexterous manipulation.
• Experience with large-scale distributed training.
• Willingness to spend time on factory and warehouse floors.

Conditions

Based in Watertown, MA, where the company is headquartered.

Field work across the United States is part of maintaining the fleet.

Equal opportunity employer that welcomes applicants from all backgrounds.

Benefits

• The company covers the full cost of your health insurance premium.
• A managed 401(k) to help you save for later.
• Unlimited PTO and all federal holidays off for full-time salaried employees.
• Hourly employees get 16+ days of accrued paid time off, plus 1.5x pay on federal holidays.
• Regular in-office meals and team social events throughout the year.
• Benefits are set to expand as the company grows, with continued investment in support for the team at work and beyond.