RL Research Scientist: Dexterous Manipulation & Sim-to-Real

Boston Dynamics; Boston (MA); On-site; USD 175,000 - 220,000; Full time

Benefits offered by this job

Medical benefits

401(k)

Paid time off

Annual bonus

Job summary

Boston Dynamics is seeking an RL Research Scientist to lead projects on visual sim-to-real transfer and post-training of Vision-Language-Action models. You will turn unlabeled data from simulation or real world into robust dexterous manipulation skills for bimanual and multi-fingered systems in unstructured environments.

You will publish in top venues and push the frontier of manipulation with tactile sensing, world models, and large-scale RL.

Qualifications

• PhD in ML, Robotics or related field or MS with 3+ years experience.
• Track record of first-author publications at top venues (CoRL, RSS, ICLR, NeuRIPS).
• Hands-on experience training policies for dexterous or contact-rich manipulation.
• Experience with VLA models, diffusion policies, or large behavior models.
• Proficient in PyTorch and/or JAX with scalable model training.
• Strong software fundamentals and ability to ship research code.

Responsibilities

• Develop novel algorithms for visual sim-to-real transfer with photorealistic rendering.
• Design post-training recipes to improve pretrained VLA models on manipulation tasks.
• Research reward modeling and offline-to-online RL for large multimodal policies.
• Close the sim-to-real gap through tactile sensing, vision, and system identification.
• Train policies that generalize across objects, scenes and embodiments.

Education

PhD or MS with 3+ years experience

Tools

PyTorch

JAX

Job description

Boston Dynamics is seeking an RL Research Scientist to lead projects on visual sim-to-real transfer and post-training of Vision-Language-Action models. You will turn unlabeled data from simulation or real world into robust dexterous manipulation skills for bimanual and multi-fingered systems in unstructured environments.

You will publish in top venues and push the frontier of manipulation with tactile sensing, world models, and large-scale RL.