PhD in ML, Robotics, or related field, or MS with 3+ years experience
Proven first-author publications in top-tier conferences (CoRL, RSS, ICLR, NeuRIPS)
Hands-on experience with dexterous or contact-rich manipulation training
Experience with VLA models, diffusion policies, or large behavior models
Proficient in PyTorch and/or JAX for training models at scale
Strong software fundamentals with ability to deliver reliable research code
Responsibilities
Develop novel algorithms for visual sim-to-real transfer using photorealistic rendering
Design post-training strategies to enhance pretrained VLA models for manipulation
Research reward modeling and offline-to-online reinforcement learning for large multimodal policies
Bridge the sim-to-real gap utilizing tactile sensing, vision, and system identification
Train robust policies that generalize across varying objects, scenes, and robotic embodiments
Benefits
Direct impact on humanoid robot policies in the real world
Access to advanced robotic fleet, simulation infrastructure, and high-performance computing resources
Opportunity to redefine capabilities in dexterous manipulation on a large scale
Full Job Description
Are you passionate about using reinforcement learning to solve dexterous manipulation tasks? As an RL Research Scientist, you'll lead research projects on visual sim-to-real transfer and post-training of Vision-Language-Action (VLA) models. Your job is to turn unlabeled data from simulation or the real-world into robust real-world manipulation skills. You should push the frontier of what bimanual and multi-fingered systems can do in unstructured environments.
In this role, you will:
Develop novel algorithms for visual sim-to-real transfer with photorealistic rendering
Design post-training recipes that improve pretrained VLA models on manipulation tasks
Research reward modeling, and offline-to-online RL for large multimodal policies
Close the sim-2-real gap through tactile sensing, vision, and system identification
Train policies that generalize across objects, scenes, and embodiments
Required Qualifications:
PhD, in ML, Robotics, or a related field or a MS with 3+ years of experience
Track record of first-author publications at top venues (CoRL, RSS, ICLR, NeuRIPS)
Demonstrated experience training policies for dexterous or contact-rich manipulation
Hands-on experience with VLA models, diffusion policies, or large behavior models
Proficient in PyTorch and/or JAX, with experience training models at scale
Strong software fundamentals and the ability to ship research code that runs reliably
The ideal candidate has:
Deployed vision-based manipulation policies on physical robots
Deep knowledge of sim-to-real transfer techniques and photorealistic rendering
Built training pipelines that combine RL, imitation learning, and large-scale pretraining
Experience fine-tuning foundation models with RLHF, DPO, GRPO, or related methods
Familiarity with tactile sensing, multi-fingered hands, or bimanual coordination
Why join us?
Direct impact on the policies powering our humanoid robots in the real world
Access to a world-class fleet of robots, simulation infrastructure, and compute
The chance to define what's possible for dexterous manipulation at scale
The pay range for this position is between $175,000 to $220,000 annually. Base pay will depend on multiple individualized factors including, but not limited to internal equity, job related knowledge, skills and experience. This range represents a good faith estimate of compensation at the time of posting. Boston Dynamics offers a generous Benefits package including medical, dental vision, 401(k), paid time off and a annual bonus structure. Additional details regarding these benefit plans will be provided if an employee receives an offer for employment.
#LI-JM1
About Boston Dynamics
Boston Dynamics is an American engineering and robotics design company founded in 1992 as a spin-off from the Massachusetts Institute of Technology. The company is best known for the development of BigDog, a quadruped robot designed for the U.S. military. Boston Dynamics has also developed a number of other robots, including Spot, a four-legged robot designed for indoor and outdoor operation, and Atlas, a humanoid robot designed for a variety of search and rescue tasks. In 2013, the company was acquired by Google X, a subsidiary of Alphabet Inc. In 2020, the company was acquired by Hyundai Motor Group. Boston Dynamics is headquartered in Waltham, Massachusetts.
