inflearn logo

Understanding Core LLM Theory Through Architecture - How ChatGPT, RAG, and Agents Work, All at Once -

You use ChatGPT, but haven’t you found it difficult to explain why it produces answers like these? “I know terms like RAG, agents, and fine-tuning… but it’s difficult to explain them precisely.” “When I hear LLM-related terminology, I’m at a loss for words.” “Explanations of concepts in AI meetings are always vague.” This course was created specifically for people like you. This is a theory course designed to help you understand LLMs not as ‘tools,’ but as ‘structures.’ Rather than teaching you how to use ChatGPT or Gemini, it gives you a framework for explaining why they work the way they do.

(4.5) 31 reviews

212 learners

Level Beginner

Course period 4 months

ChatGPT
ChatGPT
prompt engineering
prompt engineering
LLM
LLM
RAG
RAG
AI Agent
AI Agent
ChatGPT
ChatGPT
prompt engineering
prompt engineering
LLM
LLM
RAG
RAG
AI Agent
AI Agent

Reviews from Early Learners

4.5

5.0

HJ

100% enrolled

This is really great material for grasping the overall concepts! How can I receive the lecture materials?

5.0

Goldie

63% enrolled

It is good for building a solid foundation.

5.0

leckar1231

100% enrolled

It would be even better if lecture materials were provided and Q&A responses were addressed. The content is truly great for grasping the overall concepts!

What you will gain after the course

  • Structural thinking to understand the process by which an LLM generates answers

  • Criteria for not confusing key concepts such as prompts, RAG, and agents

  • An understanding that allows you to accurately follow discussions related to AI

  • Realistic judgment informed by the limitations and parameters of LLMs


The principles behind how LLMs work
A comprehensive theory course to make them your own

Gain a deep understanding of LLMs, the core technology driving the AI era.


Have you been using ChatGPT or Gemini but wondered how they work?
This course explains everything step by step, starting with the basic concepts of LLMs,

It helps you understand the core technologies of prompts, RAG, and agents without difficulty.

How LLMs WorkWinvalid

A theory course to make it your own

Although you use ChatGPT and Gemini,

Have you ever wondered why it gives answers like that?

This course covers everything from the basic structure of LLMs to their core concepts.

Explained with a focus on understanding, without complex formulas.

Transformer, Self-Attention, and more

Prompts, RAG, and Agents

You can naturally connect and understand how they work inside an LLM.

Not how to use tools, but rather,

This course helps you develop the criteria needed to evaluate AI.


This course helps you understand the thought structure of LLMs,

For leveraging the latest technologies such as prompts, RAG, and agents

This is a course that lays the foundation.

What makes this course different?

This course does not simply cover how to use tools or various tricks.

LLMs being

  • How it understands context and

  • Why hallucinations occur and

  • why Prompting, RAG, fine-tuning, and Agents emerged

We explain the theory behind the key concepts step by step, without equations and with a focus on structure.

It is structured so that you can understand core concepts such as Transformers, self-attention, tokens, and embeddings
by connecting them in an intuitive flow, rather than listing papers.


Especially recommended for those who:

  • If you use ChatGPT but are always confused about LLM concepts

  • Planners and PMs who don't understand RAG or Agent discussions in meetings

  • Business practitioners considering AI adoption or utilization strategies

  • Not a developer, but someone who wants to truly understand LLMs

  • Learners looking for an “Introduction to LLM Theory” course

This course is not this kind of course.

  • ❌ A lecture explaining ChatGPT’s features

  • ❌ A course focused primarily on how to use specific AI tools

  • ❌ Hands-on course focused on automation and work efficiency

This course is a theory-focused course that helps you understand the structure and operating principles of LLMs.

By taking this course

From someone who simply uses AI to
someone who can understand and design AI


Structure-focused learning to
understand the core principles of LLMs


Section 1 - Basic Understanding of Generative AI and LLMs

Explore the basic principles of generative AI and LLMs. Understand how LLMs statistically learn the meaning and context of language from vast amounts of text data to generate natural sentences.

Section 2 - Analysis of LLM Trends and Industry Developments

We examine the latest development trends in LLM technology and analyze future strategic directions for LLMs amid the global AI competition. This will help provide a view of the present and future of LLM technology.

Section 3 - Fundamentals of How LLMs Work

Learn the fundamental operating principles of LLMs. Understand key concepts such as tokens, embeddings, vector spaces, and context windows, and learn about the main output control parameters.

Section 4 - Prompt Engineering Techniques

Covers the basic concepts and advanced techniques of prompt engineering to maximize LLM performance. Learn various patterns, including Zero-shot, Few-shot, and Chain-of-Thought (CoT), to develop the ability to write effective prompts.

Section 5 - Addressing LLM Limitations with RAG

Learn the RAG (Retrieval-Augmented Generation) architecture to overcome limitations of LLMs, such as hallucinations and a lack of up-to-date information. Understand embeddings and vector databases, the core components of RAG, and grasp the overall workflow.

Section 6 - RAG Performance Improvement Strategies

Learn the metrics and methodologies for evaluating the accuracy of RAG systems, and explore practical techniques for improving performance. Through this, investigate ways to enhance the efficiency and reliability of RAG systems.

Section 7 - Fine-Tuning and Parameter-Efficient Tuning Strategies

Learn the basic concepts and application strategies of fine-tuning LLMs for specific tasks or domains. You will also learn efficient model tuning methods through lightweight tuning techniques such as PEFT (Parameter-Efficient Fine-Tuning).

Section 8 - Understanding and Using LLM Agents

Understand the concept and structure of Agents, and explore various types of LLM Agents. Learn through concrete examples how Agents can be applied to actual tasks and services.

Section 9 - Latest Theories on MCP and A2A

Compare and analyze the concepts, operating principles, structures, and applications of MCP (Multi-agent Cooperative Planning) and A2A (Agent-to-Agent), the latest theories in multi-agent systems. Through this, you will gain an understanding of advanced agent system design.

From Theory to Practice


Point 1. Understand the Core Principles of LLMs Without Equations

Do you use ChatGPT but wonder how it works? This course clearly explains how LLMs operate, focusing on their structure without using formulas. You’ll understand the fundamental reasons behind hallucinations, as well as why prompts, RAG, fine-tuning, and agents are necessary.


Point 2. You can easily understand how LLMs work.

Rather than merely learning how to use tools, the focus is on understanding the thought structure by which LLMs operate.

Through this, you will gain a theoretical framework for understanding and explaining concepts that arise in AI-related meetings and planning documents without confusion. In addition, by understanding LLMs’ structural limitations, such as hallucinations, recency, and context limitations,

You can determine what to expect and what not to expect.

Point 3. A theoretical framework for understanding RAG, fine-tuning, and Agents

Understand theoretically the core components and overall workflow of the RAG architecture, which was introduced to address the structural limitations of LLMs. Also, structurally explain what problems fine-tuning and lightweight tuning techniques (PEFT, LoRA) were designed to solve, and when they are appropriate choices. Agents are likewise discussed not in terms of “how to build them,” but with a focus on why they are needed, their internal structure, and how they differ from simple automation.


Point 4. Developing a perspective for understanding AI

The goal is to move beyond simply using AI and develop a structural understanding of how it works.

From core LLM concepts such as tokens, embeddings, and context windows to the latest multi-agent theories, including MCP and A2A

It provides an overview centered on why this structure emerged.


I use ChatGPT, but I don’t understand why it works the way it does.
This course was created specifically for people like you.


✔️ Beginners who want to understand the basic principles of LLMs

  • Those who want to structurally understand why LLMs hallucinate (Hallucination)

  • Those who want to understand why prompts, RAG, fine-tuning, and agents are needed and how they work

  • Those who want to fundamentally understand how AI works beyond simply learning how to use tools

✔️ Planners and practitioners who want to accurately explain LLM concepts in AI-related meetings or planning documents

  • Those who want to clearly explain the core principles of LLMs (tokens, embeddings, and context windows)

  • Those who want to understand the latest LLM technology trends, including RAG, fine-tuning, and agents

  • Those who want to explore practical ways to leverage LLMs while taking their limitations into account when planning and developing AI services and strategies

✔️ Practicing developers/data analysts who want to apply LLMs effectively to their work

  • Those who want to structurally understand LLM reasoning and answer-generation processes and apply them to development

  • Those who want to establish criteria for determining which approach (prompting, RAG, fine-tuning, or agents) is best suited to solving a problem

  • Those who want to overcome LLM limitations (hallucinations, recency, and context) and design strategies for applying them to real-world services


Stop using AI like a 'black box.'
Become an expert who understands how LLMs work.

Things to Know Before Enrolling

  • This course focuses on
    understanding the structure and operating principles of LLMs (large language models) and is a
    theory-focused course.

    • The hands-on exercises are supplementary tools to help you understand the concepts.

    • It does not aim to teach how to use specific AI tools or automate practical tasks.

Prerequisite Knowledge and Notes

  • A basic understanding of key concepts such as LLMs, Transformers, and Self-Attention is recommended.

  • Familiarity with related terms such as tokens, context windows, and embeddings will help with learning.

  • Curiosity about how AI LLMs work and a willingness to learn are important.



Recommended for
these people

Who is this course right for?

  • Core Principles of LLMs, Transformers, and Self-Attention

  • A summary of essential concepts such as tokens, context windows, and embeddings

  • Core Prompt Engineering Techniques (Zero-shot, Few-shot, CoT)

  • Overall structure of RAG and methods for improving accuracy

  • Differences Between Fine-Tuning and RAG and Criteria for Choosing Them

  • Architecture and Real-World Use Cases of AI Agents

  • The latest trends in multi-agent theory, such as MCP and A2A

Need to know before starting?

  • People who use ChatGPT but are always confused about the concept of LLMs

  • A planner/PM who doesn’t understand what’s being discussed in meetings when RAG or agents come up

  • Business practitioners considering AI adoption or utilization strategies

  • For those who aren’t developers but want to properly understand AI LLMs

  • A learner looking for an “Introduction to LLM Fundamentals” course

Hello
This is HappyAI

5,581

Learners

341

Reviews

54

Answers

4.5

Rating

12

Courses

Lee JinKyu | Lee JinKyu

Ph.D. in AI Engineering · Adjunct Professor in the Department of AI Software · CEO of Happy AI

Hello.
I am Lee JinKyu, and I research, develop, and teach AI, focusing on natural language processing (NLP) and large language models (LLMs).

I earned a Ph.D. in Engineering specializing in Artificial Intelligence and have conducted research primarily in natural language processing and LLMs.

I am currently an adjunct professor in the Department of AI Software, teaching AI-related courses such as natural language processing and computer vision, and I run HappyAI, a generative AI company.

After working as a natural language processing and big data analysis researcher at a government-funded research institute, and then as a developer at an AI-specialized company, I currently carry out AI projects and training for businesses and public institutions.

At various companies and institutions, including Samsung SDS, KT, Hyundai Wia, and the Seoul Digital Foundation, I have conducted training and projects related to generative AI, LLMs, RAG, AI agents, and fine-tuning.

Additionally, I have conducted natural language processing and text mining research on various types of unstructured data, including documents, surveys, reviews, news media, policy, and academic data, and have experience with research and analysis projects across fields such as healthcare, policy, the environment, law, economics, and education.


Key Career History

  • Adjunct Professor in the Department of AI Software

    • Natural Language Processing

    • Computer Vision

    • Generative AI

    • Data Mining

  • Ph.D. in Engineering (Artificial Intelligence)

     

    • Large Language Model (LLM) and Agent Research

  • CEO of Happy AI

    • Generative AI, LLM, RAG, and AI Agent R&D and education

  • Researcher at a government-funded research institute

    • Natural Language Processing · Big Data Analytics

  • Developer at an AI and Big Data Specialist Company

  • AI Columnist

    • Articles on generative AI, LLMs, RAG, and AI technologies


Key Areas of Expertise

  • Generative AI and large language models (LLMs)

  • RAG · Private LLM

  • AI Agent · Agentic AI

  • AI applications based on LangChain and LangGraph

  • LLM fine-tuning based on LoRA·QLoRA

  • Natural Language Processing (NLP) · Text Mining

  • AI-based data analysis and work automation


Training for Major Companies and Institutions

  • Samsung SDS – LangChain·RAG-based LLM programming

  • KT – LLM/sLLM application development

  • Seoul Digital Foundation – LLM theory and RAG chatbot development

  • Hyundai Wia – Generative AI·AX training

  • Seoul National University of Science and Technology – Python-based text analysis

  • Kyonggi University – Python Using ChatGPT

  • Dankook University – Big Data Expert Program


Key AI Projects

  • Private LLM-based RAG document search and chatbot development

  • Development of Private LLM solutions in an internal network environment

  • LLM Fine-Tuning and Instruction Tuning

  • AI Agent-based business process design

  • Natural language processing and text mining–based research and analysis

  • AI analysis of survey, review, media, policy, and academic data


Research and Publications

  • Conducted domestic and international academic research on natural language processing and LLMs

  • Research on measuring and mitigating LLM bias

  • Numerous papers in the fields of NLP and text mining

  • Experience in AI-related patents and software research and development

  • Book: “Stock Data Analysis with ChatGPT”


What I Consider Most Important in My Lectures

AI technology is changing rapidly, but if you understand the core principles, you can learn new technologies much faster.

In the lectures, rather than simply following features or code,

“what it is → why it is needed → how it works → where it is used in practice”

I consider it most important to explain things clearly so that you can understand them.

Rather than explaining complex AI technologies in a difficult way,
I will explain them so that even beginners can understand the overall structure and apply them to real-world work and projects.


Inquiries about lectures and projects

Email
leejinkyu0612@naver.com

Homepage
https://happyaidata.kr

YouTube
https://www.youtube.com/@HappyAI_0612

GitHub
https://github.com/leejin-kyu

※ Kmong Prime Expert

 

Detailed profile
https://bit.ly/jinkyu-profile

More

Curriculum

All

28 lectures ∙ (1hr 31min)

Course Materials:

Lecture resources
Published: 
Last updated: 

Reviews

All

31 reviews

4.5

31 reviews

  • leckar12311219님의 프로필 이미지
    leckar12311219

    Reviews 23

    ∙

    Average Rating 5.0

    5

    100% enrolled

    It would be even better if lecture materials were provided and Q&A responses were addressed. The content is truly great for grasping the overall concepts!

    • leejinkyu0612
      Instructor

      Thank you for the positive review. Please contact leejinkyu0612@naver.com to receive the lecture materials.

  • 20007013611님의 프로필 이미지
    20007013611

    Reviews 1

    ∙

    Average Rating 5.0

    5

    100% enrolled

    It was a great introductory session for listening to the latest key AI concepts.

    • goldie7님의 프로필 이미지
      goldie7

      Reviews 12

      ∙

      Average Rating 5.0

      5

      63% enrolled

      It is good for building a solid foundation.

      • present29263701님의 프로필 이미지
        present29263701

        Reviews 2

        ∙

        Average Rating 5.0

        Edited

        5

        100% enrolled

        This is really great material for grasping the overall concepts! How can I receive the lecture materials?

        • ymkoo님의 프로필 이미지
          ymkoo

          Reviews 10

          ∙

          Average Rating 5.0

          5

          33% enrolled

          It is easy to understand, and the clear explanations are very helpful for studying.

          HappyAI's other courses

          Check out other courses by the instructor!

          Similar courses

          Explore other courses in the same field!

          25% off for new members

          $41.70

          25%

          $51.70