PhD Research Intern, Generalist Embodied Agents Research - 2027

at Nvidia

USD 38-94 per hour

INTERN

✅ On-site

Tech Stack
Tag name is followed by "@" symbol and proficiency level value. About proficiency levels:

1-2 — basic awareness. Minimal hands-on experience, and a rudimentary understanding of the technology's purpose;

3-6 — daily use. Comfortable and regular usage, capable of handling common tasks and challenges related to the technology;

7-9 — you are an expert, you can teach others, you know all the pitfalls and tricks;

10 — exceptional knowledge, comprehensive understanding, and adeptness in all aspects of the technology, including advanced problem-solving. Think twice before claiming or demanding such level.

AI @ 5 Algorithms CUDA @ 5 JAX @ 3 LLM Machine Learning @ 3 PyTorch @ 3 Python @ 5 Reinforcement Learning @ 3 Robotics @ 3 TensorFlow @ 3

Details

NVIDIA is searching for an outstanding research intern to build humanoid robot foundation models and systems in the Generalist Embodied Agent Research (GEAR) group. The mission is to build general-purpose embodied agents that learn to explore and master complex skills across virtual and physical worlds.

You will work with a collaborative research team producing influential work on multimodal foundation models, large-scale robot learning, game AI, and physical simulation. Past projects include Eureka, VIMA, Voyager, MineDojo, MimicPlay, Prismer, and Project GR00T, a foundation model for humanoid robots. Contributions will support research projects and product roadmaps.

Responsibilities

Design and implement novel AI algorithms and models for general-purpose humanoid robots and embodied agents.
Develop large-scale AI training and inference methods for foundation models.
Optimize and deploy AI models in physical simulation and on robot hardware.
Collaborate with research and engineering teams across NVIDIA to transfer research to products and services.

Requirements

Pursuing a PhD degree in Computer Science, Computer Engineering, Electrical Engineering, or a related field.
Outstanding engineering skills in rapid prototyping and model training frameworks such as PyTorch, JAX, or TensorFlow.
Python proficiency is required; C++ and CUDA proficiency is a plus.
Excellent skills working with large-scale machine learning and AI systems and computer infrastructure.
Experience in at least one of the following areas: multimodal foundation models or robotics.
Hands-on training experience and publications involving at least one of the following: large language models, vision-language models, video generative models, diffusion models, or action-based transformers.
For robotics-focused candidates: hands-on training experience and publications in robot learning, including reinforcement learning, imitation learning, or classical control methods.
Deep understanding of robot kinematics, dynamics, and sensors.
Ability to safely operate robot hardware, lab equipment, and tools.
Knowledge of control methods such as PID, MPC, and whole-body control.
Familiarity with physics simulation frameworks such as MuJoCo and the Isaac suite.

Research Publications

GEAR publications include:

DreamDojo: A Generalist Robot World Model from Large-Scale Human Videos
EgoScale: Scaling Dexterous Manipulation with Diverse Egocentric Human Data
DreamZero: World Action Models are Zero-shot Policies
SONIC: Supersizing Motion Tracking for Natural Humanoid Whole-Body Control

Compensation and Benefits

The internship hourly rate is USD 38–94, based on position, location, year in school, degree, and experience. Interns are also eligible for NVIDIA intern benefits.

Applications will be accepted at least until September 21, 2026. This posting is for an existing vacancy. NVIDIA uses AI tools in its recruiting processes and is an equal opportunity employer.