Mandatory Prerequisites
Languages/Tools
Python, Linux or Windows 10/11, Docker/containers, Git, basic shell commands, Jupyter/VS Code, familiarity with 3D concepts (meshes, materials, coordinate transforms).
Frameworks
JetPack, CUDA, NGC containers, TensorRT, NVIDIA Triton Inference Server™, DeepStream, Isaac ROS / ROS 2, NVIDIA Isaac Sim™, Kit App Template, NVIDIA Omniverse™ Kit, OpenUSD, Nucleus (optional), OmniGraph; optional VLM/VSS workflows and Python extension scripting experience.
Bring to the Workshop
Laptop with internet access. Ideal minimum: 10 Mbps download / 2 Mbps upload. Brev cloud credits provided.
Optional
Bring CAD files (JT, FBX, STEP, OBJ), USD assets, ROS bags, sample videos, model artifacts, robot logs, or project-specific requirements.
Agenda
Jetson Orin, Jetson Thor, NVIDIA Cosmos™ (reason/predict/transfer), GR00T, Isaac Sim, Isaac Lab, DeepStream, TRT, Triton, VSS, and open model stack
Live 3-computer demo: Jetson Thor (edge deployment), NVIDIA RTX PRO™ 6000 (simulation), and VLM/VLA hardware-in-the-loop
Cosmos fine-tuning, VLM/VLA development, sensor input via browser mic/camera, edge deployment walkthrough (same pipeline as the Thor showcase)
Isaac Lab-Arena setup, demo data conversion (HDF5 → LeRobot), GR00T 1.7 VLA post-training, closed-loop pick-and-place evaluation (on Brev)
Speaker
Senior Solutions Architect
NVIDIA
Ninad Madhab specializes in edge AI, robotics, and industrial AI deployment using platforms such as Jetson, DeepStream, Isaac, and Metropolis Microservices. He brings hands-on experience in Jetson-based development, helping teams accelerate edge AI rollout and move AI applications from model development to real-world factory floor deployment.
Event Details
NVIDIA Physical AI Hands-On Training
October 7, 9:00 a.m. – 5:00 p.m.
WeWork EON Free Zone
Kharadi, Cluster D, Wing 4, EON Free Zone, Kharadi, Pune, MH 411016
Venue
WeWork EON Free Zone
Kharadi, Cluster D, Wing 4, EON Free Zone, Kharadi, Pune, MH 411016
Get Started With Your Open Source AI Journey
Discover how Sarvam is partnering with NVIDIA to architect India’s sovereign AI infrastructure with NVIDIA’s full hardware and software stack. By training its foundational models from scratch on NVIDIA H100 GPU clusters using the NVIDIA NeMo™ framework and Megatron-LM, Sarvam is natively processing over 2 trillion authentic Indian language tokens.
Using open technologies like the NVIDIA Nemotron™ toolchains has enabled Sarvam to scale securely within national trust boundaries.
Additional Resources
NVIDIA Jetson Kits |
Base Jetson software stack: Jetson Linux, CUDA, TensorRT, cuDNN, VPI, multimedia APIs. |
|
NVIDIA SDK Manager |
Flashing Jetson and installing JetPack / host-target setup |
|
NVIDIA Jetson Platform Services |
Modular, API-driven, cloud-native services for deploying edge AI / GenAI apps faster on Jetson. |
|
NVIDIA Jetson AI Lab |
Hands-on GenAI examples: LLMs, VLMs, diffusion models and tutorials |
|
NVIDIA Isaac ROS Or DeepStream |
Isaac ROS for robotics; DeepStream for video analytics / IVA. |
|
NVIDIA Developer Program |
Access free tools, extensive learning opportunities, and expert help with the NVIDIA Developer Program. |
|
NVIDIA Jetson Kits |
About NVIDIA Jetson Kits |
|
NVIDIA Jetson Tech Blog |
Maximizing Memory Efficiency to Run Bigger Models on NVIDIA Jetson |
|
OpenUSD |
Universal Scene Description (OpenUSD) is an open and extensible industry standard 3D format used for composing, simulating, and collaborating within 3D worlds |
|
SimReady |
SimReady (simulation-ready) is the standard specification for 3D assets that incorporate real-world physical properties, behaviors, semantic labeling, and metadata. |
|
NVIDIA Issac Groot |
NVIDIA Isaac™ GR00T is an open reference platform for general-purpose humanoid robots that enables developers to more efficiently build, train, test, and deploy AI-powered robots |
|
NVIDIA Issac Groot |
Accelerating Humanoid Robot Development With NVIDIA Isaac GR00T |
|
NVIDIA Issac Groot |
Isaac Groot Documentation |
|
NVIDIA Cosmos |
Develop physical AI faster with leading world foundation models and open data processing, training, and evaluation frameworks |
|
NVIDIA Cosmos |
NVIDIA Cosmos Github |
|
NVIDIA Cosmos |
Introducing NVIDIA Cosmos 3: The Open Model That Thinks, Generates, and Acts |