Building Agentic AI Applications with LLMs
NVIDIA Deep Learning Institute
Last updated June 18, 2026
Building Agentic AI Applications with LLMs is an eight-hour, hands-on course from NVIDIA's Deep Learning Institute for developers who want to build systems where language models do more than answer questions — they plan, call tools, retrieve information, and coordinate with one another to complete larger tasks. You begin with agent fundamentals and the strengths and limits of LLMs, then learn to constrain a model into structured, machine-readable outputs so its responses can drive function calls and API integrations. From there the course moves through retrieval mechanisms and knowledge graphs for grounding agents in domain knowledge, and into its central subject: multi-agent systems, where specialized agents are decomposed, given communication channels, and run concurrently using orchestration frameworks — LangGraph in particular. It closes with a final assessment in which you deploy an agent that schedules multiple retrieval operations and reports back. The work is in Python and PyTorch and uses NVIDIA NIM alongside LangChain and LangGraph.
What you'll learn
- Building a minimal agent and reasoning about where LLMs help and where they fail
- Forcing structured, schema-stable outputs so a model can drive function calls and API integrations
- Using retrieval and knowledge graphs to ground agents in domain knowledge
- Decomposing work across specialized, concurrent agents and orchestrating them with LangGraph
Frequently asked questions about Building Agentic AI Applications with LLMs
Who is Building Agentic AI Applications with LLMs for?
Developers with intermediate Python and introductory deep-learning background (including attention and transformers) who want to design multi-agent systems and orchestrate them in real software.
Is Building Agentic AI Applications with LLMs free?
No — Building Agentic AI Applications with LLMs is a paid course.
What are the prerequisites for Building Agentic AI Applications with LLMs?
Introductory deep-learning knowledge, including attention mechanisms and transformers (the experience from DLI's Getting Started with Deep Learning or Fundamentals of Deep Learning is preferred). Intermediate Python proficiency, including object-oriented programming and familiarity with ML libraries, is also expected.
Do you need to code for Building Agentic AI Applications with LLMs?
Yes — Building Agentic AI Applications with LLMs involves hands-on coding.
Why we suggest this course
This course overlaps deliberately with our other agent and RAG titles on the fundamentals — agents, tool use, retrieval — so its reason to exist on the track is the part those others don't center on: multi-agent orchestration with LangGraph and the structured-output discipline that makes agents reliable enough to wire into real software. The distinct takeaway is hands-on practice decomposing a task across concurrent, communicating agents rather than building a single one. If you only want one agent-building course, "Build a Deep Research Agent" is the more focused single-system project; choose this one when the multi-agent and orchestration angle is what you're after.