Lý thuyết
Bài 5/5

AI Agents & LangGraph

Xây dựng AI Agents tự động - Autonomous systems có thể reasoning và sử dụng tools.

📋 Mô tả khóa học

AI Agents là future của AI applications - systems có thể tự suy nghĩ, plan, và execute tasks phức tạp. Khóa học này giúp bạn build agents với LangGraph, từ simple tool-calling đến multi-agent systems.

🎯 Bạn sẽ học được gì?

  • ✅ AI Agent architecture (ReAct, Plan-and-Execute)
  • ✅ LangGraph for agent workflows
  • ✅ Tool calling & function execution
  • ✅ Multi-agent systems & collaboration
  • ✅ Memory & state management
  • ✅ Agent evaluation & monitoring

👥 Khóa học dành cho ai?

  • Developers building autonomous AI
  • Engineers creating AI assistants
  • Teams implementing AI automation
  • Anyone wanting cutting-edge AI skills

📚 Chương trình học (16 bài)

Module 1: Agent Fundamentals

  1. What are AI Agents - Definition, capabilities, limitations
  2. ReAct Pattern - Reasoning + Acting
  3. LangGraph Basics - Nodes, edges, state
  4. First Agent - Simple tool-using agent

Module 2: Tool Use & Functions

  1. Tool Calling - OpenAI function calling
  2. Building Custom Tools - Python functions as tools
  3. Tool Selection - When to use what tool
  4. Error Handling - Graceful failure, retries

Module 3: Advanced Agents

  1. Memory Systems - Short-term, long-term memory
  2. State Management - Checkpointing, persistence
  3. Planning Agents - Plan-and-execute pattern
  4. Self-Reflection - Self-correction, improvement

Module 4: Multi-Agent Systems

  1. Multi-Agent Architecture - Supervisor, hierarchical
  2. Agent Communication - Handoffs, collaboration
  3. Production Agents - LangSmith monitoring
  4. Capstone Project - Complete agent system

🛠️ Tech Stack

  • Python 3.10+
  • LangGraph, LangChain
  • OpenAI API, Anthropic API
  • LangSmith for monitoring
  • FastAPI for deployment

🚀 Dự án chính

  1. Research Assistant Agent - Web search, summarize, report
  2. Data Analysis Agent - Query data, visualize, insights
  3. Multi-Agent Customer Support - Routing, specialists, escalation
  4. Code Review Agent - Analyze PRs, suggest improvements

⚙️ Prerequisites

  • ✅ GenAI Image & Multimodal course
  • ✅ Python advanced level
  • ✅ Understanding of async programming

Thời lượng: 8-10 tuần (6-8 giờ/tuần)
Level: Advanced
Pathway: GenAI Coding

Bắt đầu học →