Production-Grade Agentic GenAI Systems
Moving beyond simple prompts to autonomous, self-correcting AI workflows.
Master the transition from static LLM calls to autonomous agents that can reason, use tools, and collaborate. This course focuses on the engineering rigor required to move agentic workflows from 'cool demo' to 'production-ready infrastructure'.
- Level
- Advanced
- Status
- Available
- Duration
- 8 Weeks
- Instructor
- Hoan Do (Master's from EPFL)
Curriculum
The Agentic Shift
Duration: 1h 30m
- From Chains to Agents — 20 min
- The Cognitive Architecture (ReAct) — 30 min
- Lab: Building a ReAct Loop from Scratch — 40 min
State & Memory (LangGraph)
Duration: 2 Hours
- Graphs vs. Chains — 30 min
- Persistence & Time Travel — 45 min
- Project: Customer Support Triage Bot — 45 min
Multi-Agent Orchestration (AutoGen)
Duration: 1h 45m
- The Team Pattern — 25 min
- AutoGen & Conversation Architectures — 40 min
- Pattern: The "Code Execution" Sandbox — 40 min
Production Challenges
Duration: 1h
- The Hallucination Loop — 20 min
- Observability (LangSmith) — 25 min
- Evaluation (LLM-as-a-Judge) — 15 min