About This Course
Welcome to the "Al Agent - Building Agents with LangGraph and A2A" course, an intermediate-level, hands-on program for software engineers, embedded systems developers, and tech consultants who already understand LM basics and want to move from building linear chatbots to constructing complex, autonomous systems. It is ideal for professionals implementing controllable, multi-agent automated systems within enterprise or industrial environments.
Over two days you will master the architectural principles and practical implementation of stateful, multi-agent systems with LangGraph. You will design, build, debug, and connect production-grade agents that handle tools, long-term memory, human approval gates, and autonomous agent-to-agent (A2A) handoffs - tracing and evaluating them with LangSmith - and finish with a capstone where you build and present a working prototype agent.
Learning Objectives
Upon completing this course, you will be able to:
- Explain agent engineering architecture and its key building blocks: LLM, tools, state, memory, routing, workflow, and orchestration.
- Distinguish agent design patterns: single-agent, workflow-agent, supervisor-agent, and multi-agent.
- Build controllable agent workflows in LangGraph using nodes, edges, state, graph execution, and conditional routing.
- Design and integrate tools with function calling and structured execution, handling tool errors and unsafe requests.
- Manage short-term state and long-term memory, with checkpoints, persistence, and recovery.
- Add human-in-the-loop approval gates before sensitive actions.
- Design multi-agent workflows using role-based agents and supervisor/routing patterns.
- Integrate independent agents using the Agent-to-Agent (A2A) protocol with secure handoffs.
- Debug, trace, and evaluate agents using LangSmith.
- Deliver an end-to-end capstone: design, build, debug, and present a prototype agent.
Prerequisites
- Own laptop with sufficient RAM and disk space, preinstalled with Python, Node.js, Git, and Docker Desktop for the hands-on labs.
- Some signups for online tools will be required, using a Google account.
- Solid understanding of LLM basics (prompting, tokens, function calling) - this is an intermediate-level course.
- Working knowledge of Python (functions, classes, JSON) and comfort calling APIs.
- Prior experience building or using basic chatbots or LLM applications is helpful.
Target Audience
- Software engineers ready to move from linear chatbots to complex, autonomous agent systems.
- Embedded systems developers building controllable automation for industrial environments.
- Tech consultants implementing multi-agent systems in enterprise settings.
- Developers who already understand LL basics and want hands-on LangGraph and A2A experience.
- Technical leads designing stateful, production-grade agent architectures.
Training Outline
- Agent Engineering Architecture
- LangGraph Fundamentals
- Tool-Using Agents
- State, Memory, and Human-in-the-Loop
- Multi-Agent Workflow Design with LangGraph
- A2A for Agent-to-Agent Integration
- Debugging, Tracing, and Evaluation with LangSmit
- Capstone - Build and Present a Prototype Agent