below marketmost similar roles pay hereabove market
This role pays more than
52%
of similar roles. Most pay
$180,000–$254,750
— the blue band above.
At the midpoint, this role pays about
$216k
versus about
$217k
for comparable roles.
Based on 240 similar postings.
Employer
About Nvidia
Nvidia is a leading designer of graphics processing units (GPUs) and system-on-chip units, powering gaming, professional visualization, data centers, and artificial intelligence workloads. Industry: Semiconductors & AI Computing
Nvidia currently has
1623 open roles
on FindRole.
Listed pay typically runs
$184,000–$287,500
across 1110 roles with salary data.
JOB TITLE: Research Scientist, Physical AI - Foundation Models - PhD New College Grad 2026
The Research Scientist, Physical AI - Foundation Models - PhD New College Grad 2026 joins the Cosmos team to develop next-generation algorithms bridging virtual and physical realms. This role involves pioneering generative AI algorithms for physical AI applications, specifically focusing on advanced video generative models and video-language models. Key responsibilities include architecting data processing pipelines for premium training data, designing physics simulation algorithms, and scaling large-scale training systems using 20,000+ GPUs. The position requires deep expertise in PyTorch, diffusion models, vision language and reasoning models, reinforcement learning, and 3D generative models. The work addresses the technical challenge of integrating physics simulations with AI systems to create foundation models that enable digital intelligence to interact with physical reality through research-to-product transitions.
What does a Research Scientist earn?
Median $216250 from 45 postings across 16 companies.
Generative AI
PyTorch
Python
Multimodal Learning
Reinforcement Learning
Computer Vision
Distributed Training
World Models
Robotics
Simulation
Video Understanding
VLMs
Foundation Models
Action-conditioned Simulation
Policy Learning
Python
C++
PyTorch
CUDA
Reinforcement Learning
Deep Learning
Robotics
Autonomous Vehicles
Physical AI
Parallel Programming
Simulation
Closed-loop Training
Verification and Validation
Embodied AI
Multimodal Foundation Models
LLMs
Large Vision-Language Models
Diffusion Algorithms
Action-based Transformers
PyTorch
Jax
TensorFlow
Python
C++
CUDA
Reinforcement Learning
Imitation Learning
ROS
Robot Kinematics
Model Predictive Control
MuJoCo
Isaac Sim
Whole-body Control
Generative AI
PyTorch
Python
Multimodal Learning
Reinforcement Learning
Computer Vision
Distributed Training
World Models
Robotics
Video Understanding
Simulation
VLMs
Foundation Models
Action-conditioned Simulation
Policy Learning