Senior Research Engineer For Reinforcement Learning

Details

NVIDIA is searching for a senior or principal engineer specializing in large-scale reinforcement learning and policy learning in the Generalist Embodied Agent Research (GEAR) group. This team leads Project GR00T, NVIDIA's ambitious initiative to build foundation models and full-stack technology for humanoid robots.

You will collaborate with a highly skilled research team producing influential work on multimodal foundation models, large-scale robot learning, embodied AI, and physics simulation. Past projects include Eureka, VIMA, Voyager, MineDojo, MimicPlay, Prismer, and more. Your contributions will impact research projects and product roadmaps significantly.

Responsibilities

• Develop a large-scale reinforcement learning training framework capable of running on thousands of GPUs.
• Build and optimize simulation infrastructure utilizing GPU-accelerated simulators like Isaac Lab to support training locomotion and manipulation policies for robots at scale.
• Develop sim-to-real transfer pipelines and work closely with the robotics team to deploy to physical robots.
• Propose scalable solutions that combine large language models (LLMs) with policy learning (e.g., Eureka).
• Apply reinforcement learning to fine-tune multimodal LLMs.

Requirements

• Bachelor's degree or above in Computer Science, Robotics, Engineering, or related field.
• 5+ years of industry experience in large-scale deep learning or MLOps.
• Exceptional engineering skills in building, testing, and maintaining scalable distributed GPU training frameworks.
• Proficiency in Python and hands-on experience with PyTorch, JAX, or Tensorflow for model training.
• Deep familiarity with reinforcement learning algorithms such as PPO, SAC, or Q-learning, including hyperparameter tuning and reward function design.
• Familiarity with policy learning techniques like reward shaping, domain randomization, and curriculum learning.
• Strong experience with large-scale GPU clusters, HPC environments, and job scheduling/orchestration tools such as SLURM and Kubernetes.

Ways To Stand Out

• Master's or PhD degree in Computer Science, Robotics, Engineering, or related field.
• Demonstrated experience transferring policies from simulation to real robots for locomotion and manipulation.
• Contributions to popular open-source reinforcement learning frameworks or publications at top AI conferences such as NeurIPS, ICRA, ICLR, CoRL.
• Strong ability to mentor junior engineers/researchers and lead technical projects from conception to completion.

NVIDIA is known as one of the world's most desirable tech employers with some of the most forward-thinking and productive people. Join to be at the forefront of general-purpose robots and large-scale foundation model development.

Salary and Benefits

The base salary range is 148,000 USD to 287,500 USD. Salary is set based on location, experience, and pay for similar roles. Eligibility for equity and benefits is included. NVIDIA accepts ongoing applications.

NVIDIA is committed to fostering a diverse and inclusive work environment and is an equal opportunity employer.