Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Join this interactive livestream to explore training AI agents for command-line tasks using synthetic data and reinforcement learning techniques. Follow along as an NVIDIA AI research engineer demonstrates in real-time how to teach a BASH agent to use the LangGraph CLI tool, working through challenges and debugging processes as they occur. Discover how to leverage Nemotron models with NeMo Data Designer for synthetic data creation, deploy agents to reinforcement learning environments using NeMo Gym with Unsloth GRPO, and implement newly acquired skills in practical scenarios. Learn about built-in safety measures including human-in-the-loop command approval and runtime isolation that ensure secure agent operation. Understand how combining synthetic data generation with Reinforcement Learning with Verifiable Rewards (RLVR) and GRPO techniques can accelerate agent specialization for proprietary CLI environments while maintaining accuracy and trust. Participate in live Q&A sessions throughout the 51-minute session to ask questions about workflow optimization, model customization strategies, and synthetic data generation approaches for your own command-line environments.
