inflearn logo

Learning AI Agents in Practice, Part 4

Covers the practical aspects of multi-agent systems, from designing single and multi-agent systems to building collaborative architectures and enabling asynchronous communication. Learn how to use LangGraph workflows and message-broker-based parallel communication to divide complex business domains and coordinate tasks. You’ll also develop the skills to objectively evaluate and manage agent performance by building LLM-based automated scoring systems using evaluation metrics such as precision and recall, along with TDD-based quality gates.

1 learners are taking this course

Level Intermediate

Course period 12 months

Llama
Llama
LangGraph
LangGraph
TDD
TDD
multi-agent
multi-agent
Llama
Llama
LangGraph
LangGraph
TDD
TDD
multi-agent
multi-agent

What you will gain after the course

  • You can build a multi-agent collaboration system and an asynchronous parallel communication architecture using LangGraph and a message broker.

  • You can design specialist and supervisor agents tailored to the business domain and apply coordination strategies such as democratic and actor-critic approaches.

  • By building a TDD-style evaluation dataset and an LLM-based automated scoring system, we can objectively assess agent performance and manage quality gates.

Learn AI Agents Part 4 Through Real-World Practice
The era of agents working alone is over. Now make them work as a 'team'.
From single- and multi-agent design, LangGraph workflows, and asynchronous parallel communication to LLM-based automated grading and TDD quality gates.
실무에서 배우는 AI Agents Part.4 강의 대표 이미지
ABOUT THIS COURSE
You’ll be able to design, build, and evaluate multi-agent systems that break complex business domains down among multiple expert agents coordinated by a supervisor.
TOOLS

Tools Covered

These are the main tools and frameworks covered in the course.

MA

Multi-agent

LG

LangGraph

TD

TDD

LE

LLM Evaluation

OUTPUT

What You'll Build

By the end of the course, you’ll be able to design and implement the following deliverables yourself.

🧩

Domain-Partitioned Multi-Agent Design Proposal

An architecture that divides business domains into specialist agents, with a supervisor coordinating them.

Lessons 2–4
🔀

LangGraph-Based Workflow

A collaborative pipeline connecting a supervisor and specialist agents with LangGraph

Lesson 3
⚡

Parallel Communication Architecture Using a Message Broker

An implementation of parallel message processing between agents using an asynchronous queue

Lesson 5
📊

LLM-based automated grading system

A scoring pipeline that automatically measures agent performance using evaluation metrics such as precision and recall.

Lesson 6
🧪

TDD-Style Evaluation Dataset & Quality Gates

Component-level evaluation datasets and quality validation gates that must be passed before deployment

Lessons 7–8
CURRICULUM

Here’s what you’ll learn.

When grouped by topic, the curriculum builds the following skills step by step.

Lessons 1–2

Design Sense for Scaling from Single-Agent to Multi-Agent

Understand the limitations of a single agent and learn design principles for dividing a business domain into specialist units.

Lessons 3–4

Supervisor and Coordination Strategies

Implement multi-agent collaboration by comparing and applying LangGraph-based workflows and coordination strategies such as democratic and actor-critic approaches.

Lesson 5

Asynchronous and Parallel Communication Architecture

Configure parallel communication between agents using message brokers and asynchronous queues, and design an architecture that improves processing efficiency.

Lessons 6–8

Evaluation Metrics and Quality Gates

Objectively manage agent performance with precision- and recall-based automated LLM grading, TDD-style evaluation datasets, and component-level quality gates.

NOTICE

Things to Know Before Enrolling

  • This course is intended for students who have already studied the fundamentals of AI agent development, such as single agents, prompt design, and tool calling.
  • This is a self-paced course that does not provide answers to questions.
  • For effective learning, we recommend watching at normal speed and following along with the exercises rather than increasing the playback speed.
  • The tools and library versions may differ from those available at the time of recording, and major changes will be covered in the course.
  • Set up the hands-on environment on your own local or cloud system; no separate environment is provided.

Recommended for
these people

Who is this course right for?

  • Developers who want to move beyond the limitations of a single agent and adopt a multi-agent system

  • Practitioners seeking to build an objective performance evaluation and automated testing (TDD) environment for AI agents.

  • An AI service developer looking to design an architecture for collaboration among expert agents by breaking down a complex business domain.

Need to know before starting?

  • Basic Python programming and experience using LLM APIs

  • Basic understanding of AI agent frameworks such as LangChain and hands-on experience implementing a single agent

  • A basic understanding of asynchronous programming and message brokers

Hello
This is usefulit

8,281,671

Learners

6,695

Reviews

4.6

Rating

380

Courses

I will help you grow through useful IT lectures.

Reviews

Not enough reviews.
Please write a valuable review that helps everyone!

usefulit's other courses

Check out other courses by the instructor!

Similar courses

Explore other courses in the same field!

25% off for new members

$18.70

25%

$18.70