11-768 AI Agents
12 units, Carnegie Mellon University
"This course focuses on LLM-based AI agents: autonomous systems that use large language models to perceive, reason, plan, and act in complex environments. These systems represent a shift toward multi-step task completion requiring sustained interaction with external environments. LLM agents have already demonstrated a substantial impact on software development through code generation, and are poised to impact other digital domains. Developing capable agents requires specializing standard LLM training methods, including supervised fine-tuning and reinforcement learning, to the agentic domains. These systems face unique research challenges such as environmental grounding and long-horizon credit assignment, which provide significant opportunities for technical exploration. This course covers foundational capabilities like instruction following, tool use, memory, and task decomposition alongside practical considerations such as safety sandboxing, credentialing, and the use of modern development frameworks. Students will learn these established foundations as well as explore open research frontiers: advanced search techniques, and interaction with people and other agents. This course is intended for graduate students. Familiarity with training neural language models is required.
Prerequisites: none listed
Offered: Spring 2027, Fall 2026
Courseweave needs JavaScript to show sections, workloads and planning tools.