AI Agent Engineering

Frameworks, Harnesses, and Best Practices

LangChain
LlamaIndex
MCP
Agent Harness
Claude Code
Codex

Course Overview

AI Agent Engineering is transforming how we build intelligent systems. When ChatGPT writes code, customer service bots book meetings, or autonomous systems make decisions, there's sophisticated engineering underneath: AI agents that reason, use tools, remember context, and orchestrate complex workflows.

This course takes you from agent fundamentals to production-ready harnesses. You'll master the core frameworks powering modern AI systems—LangChain for agent orchestration and LlamaIndex for retrieval-augmented generation. Through hands-on labs, you'll build real agents, understand production patterns, and learn from case studies like Claude Code and Codex.

Why AI Agent Engineering Matters: As AI systems move from research to production, engineering robust, scalable, and safe agent architectures becomes critical. This course equips you with the technical skills to design multi-agent systems, implement safety controls, manage token budgets, and deploy production harnesses that power real-world applications.

You'll gain practical knowledge of agent architectures, frameworks, and production deployment patterns—no prior AI experience required.

Who Should Attend

No Technical Background Required! This course is designed for anyone interested in understanding and building AI agent systems. Whether you're a developer, product manager, entrepreneur, researcher, or simply curious about AI engineering, you'll gain practical knowledge to work with production AI agents.

Perfect For

  • • Software Engineers & Developers
  • • Product Managers & Technical Leaders
  • • Entrepreneurs & Startup Founders
  • • ML Engineers & Data Scientists
  • • Researchers & Students
  • • Anyone Curious About AI Systems

What You'll Need

  • No coding or ML background required
  • • Curiosity about AI systems and intelligent applications

Course Contents

Three comprehensive modules covering foundations to production deployment

Module 1

Foundations & Frameworks

Build your first AI agent with LangChain

Agent Architecture Fundamentals

  • • Tools, memory, planning loops
  • • ReAct pattern: Reason → Act → Observe
  • • Function calling and tool schemas

LangChain Deep Dive

  • • Chains, agents, LCEL (LangChain Expression Language)
  • • Memory modules: buffer, summary, vector store
  • • Built-in tools vs. custom tool creation
🧪 Hands-On Lab 1
Build a multi-turn research agent with web search + summarization chain and ConversationBufferMemory
Module 2

Advanced Frameworks & Components

LlamaIndex for retrieval and Model Context Protocol

LlamaIndex for Retrieval-Augmented Agents

  • • Data connectors: PDF, Markdown, GitHub, SQL
  • • Indexing strategies: VectorStore, Tree, List
  • • Query engines: retrieval, sub-question, router

Advanced Agent Components

  • • Skills/tools system: schemas, validation, error handling
  • • Memory architectures: semantic, episodic
  • • MCP (Model Context Protocol): tool discovery patterns
🧪 Hands-On Lab 2
Build a document Q&A agent: index technical documents, implement semantic search, add VectorStoreMemory
Module 3

Production Harnesses & Best Practices

Real-world case studies and harness design

Multi-Agent Orchestration

  • • Sub-agents: specialized tasks, parallel execution
  • • Delegation patterns: sequential, parallel, hierarchical
  • • Token budget management and cost control

Production Engineering Patterns

  • • Safety: sandboxing, permission systems
  • • Observability: logging, tracing, cost tracking
  • • State management: checkpointing, recovery

Case Studies

  • • Claude Code: workflow orchestration, MCP integration
  • • Codex: multi-model routing, cost optimization

Key Frameworks

LangChain

Agent orchestration framework for building multi-step reasoning systems. Covers chains, agents, and memory modules for production deployments.

Topics: LCEL (LangChain Expression Language) | ReAct Agents | Memory Systems | Tool Integration

LlamaIndex

Retrieval-augmented generation framework for building document Q&A systems. Covers data connectors, indexing strategies, and query engines.

Topics: VectorStore Indexing | Query Engines | Multi-Document Reasoning | Semantic Search

Model Context Protocol (MCP)

Standard protocol for tool discovery and integration in agent systems. Covers dynamic tool registration and server-client architecture patterns.

Topics: Tool Discovery | MCP Servers | Dynamic Registration | Resource Management

Claude Code

Production agent harness demonstrating workflow orchestration and permission systems. Case study of real-world multi-agent architecture.

Topics: Workflow Orchestration | Permission Systems | Sub-Agent Patterns | State Management

Codex

Multi-model routing and cost optimization strategies for production agent systems. Case study of deployment at scale.

Topics: Multi-Model Routing | Cost Optimization | Token Budget Management | Production Deployment

AI Agent Engineering

Frameworks, Harnesses, and Best Practices

LangChain
LlamaIndex
MCP
Agent Harness
Claude Code
Codex