Diving Deep on Physical AI | S1E2 | Fine-tuning NVIDIA GR00T on AWS

Foundation models like NVIDIA GR00T provide powerful starting points for humanoid robotics—but customization is essential for real-world applications. Learn how to fine-tune GR00T and other robotics foundation models at scale using AWS Batch and Amazon SageMaker. In this session, discover distributed training strategies, hyperparameter optimization techniques, and best practices for adapting general-purpose models to specialized tasks across manufacturing, warehouse automation, and service robotics. What you'll learn: NVIDIA GR00T foundation model architecture Distributed fine-tuning on AWS Batch with GPU instances Training workflow orchestration and monitoring Model versioning and experiment tracking Perfect for ML engineers, robotics developers, and AI practitioners building specialized humanoid robotics applications