Target Audience

• Python Developers
• AI Engineers
• Machine Learning Engineers
• Software Engineers working with LLM applications

Minimum Background Required

To gain maximum value from this workshop, participants should have:

• Good proficiency in Python
• Familiarity with Large Language Models (LLMs)

Workshop Overview

This session teaches you how to design, build, and operationalize production-ready AI agents using LangChain, LangGraph, and LangSmith. Rather than focusing on isolated prompts, you’ll learn how to architect stateful, tool-enabled systems that can reason, act, and execute multi-step workflows reliably.

Starting with core agent fundamentals, we’ll progressively move into advanced orchestration patterns using LangGraph for deterministic control and complex branching logic. You’ll implement tool calling, memory, and retrieval, then instrument your agents with LangSmith for tracing, debugging, and evaluation.

By the end of the workshop, you will have built a real-world, multi-step AI agent, understood common failure modes, and gained the practical skills required to move from prototype to production-grade systems.

Topics

• Modern LLM Application Architecture: Chains vs Agents vs Graphs
• Core Components of LangChain (Models, Prompts, Tools, Runnables)
• Tool Calling and Structured Output Patterns
• Building a Basic Tool-Using Agent
• Memory and Retrieval-Augmented Generation (RAG) for Agents
• Connecting External APIs and Custom Tools
• Introduction to LangGraph: Stateful Graph Orchestration
• Conditional Routing and Multi-Step Agent Workflows
• Human-in-the-Loop and Control of Agent Execution
• Observability and Tracing with LangSmith
• Evaluating and Testing Agents with LangSmith
• Debugging, Guardrails, and Production Readiness