Physical AI Researcher

Overview

NIST is investigating the performance of commercial and custom AI systems (hardware and software) for
Physical AI applications, specifically in machine learning, reinforcement learning, and foundation model
training for Vision-Language Models (VLMs) and Vision-Language Action (VLA) architectures. The work
will focus on leveraging physics-based simulations (e.g., Gazebo, MuJoCo, Drake) and synthetic data
generation within platforms like NVIDIA IsaacSim / GR00T / Cosmos to conduct experiments, analyze
system capabilities, and generate high-quality data to train robust Physical AI systems.

Qualifications

• Education: Engineering / Computer Science majors with Master’s Degree or Ph.D, or in the final year of degree (e.g., Computer Science, Robotics, Mechanical Engineering or similar)
• Strong programming experience in Python and C++
• Experience in AI / Machine Learning frameworks (e.g., TensorFlow, PyTorch, etc.)
• Experience with developing and applying tools for Machine Learning, Reinforcement Learning, and Foundational Models for Vision-Language Models (VLM) and Vision-Language Action (VLA) architectures
• Experience in Synthetic data generation in and for NVIDIA IsaacSim, GR00T, and Cosmos environments
• Experience in Physics-based simulation engines (e.g., Gazebo, MuJoCo, Drake, etc.)
• Experience with version control software and workflow (e.g., Git, GitHub, GitLab, BitBucket, etc)
• Experience with Unix / Linux Operating systems
• Experience with ROS 2 on Linux systems
• Some experience in with CAD software (e.g., SolidWorks, OnShape, etc.)

Research Proposal

Key responsibilities will include but are not limited to:

• Develop Physical AI applications using machine learning and reinforcement learning.
• Research and develop Foundational model training for Vision-Language Models (VLMs) and Vision-Language Action (VLA) architectures.
• Research and develop methods for synthetic data generation within platforms like NVIDIA IsaacSim / GR00T / Cosmos to conduct experiments to analyze system capabilities, and generate high-quality data to train robust Physical AI Systems.
• Write reports and develop weekly presentations to explain the progress of the project
• Work Schedule: On-campus (Gaithersburg, MD), Full-Time (40 hrs / week)