Udemy
    •  
    •  
    •  
    •  
    •  
    •  
    •  
    •  
Turn what you know into an opportunity and reach millions around the world.
Learn More
Your cart is empty.
Keep shopping
【AI 자막】 LLM 엔지니어링 : AI 인공지능 및 대규모 언어 모델 및 에이전트 마스터하기!
Highest Rated
Rating: 4.5 out of 5(163 ratings)
1,368 students

【AI 자막】 LLM 엔지니어링 : AI 인공지능 및 대규모 언어 모델 및 에이전트 마스터하기!

8주 만에 LLM 엔지니어 되기: 8개의 LLM 애플리케이션을 구축하고 배포하며 생성형 AI와 핵심 이론 개념을 마스터합니다.
Created byLigency ​, Ed Donner
Last updated 1/2025
Korean
Korean [Auto],

What you'll learn

  • 프로젝트 1: 기업 웹사이트를 지능적으로 스크래핑하고 탐색하여 AI 기반 브로셔 생성기를 제작합니다.
  • 프로젝트 2: UI 및 기능 호출을 지원하는 항공사용 멀티모달 고객 지원 에이전트를 구축합니다.
  • 프로젝트 3: 오픈 소스 및 폐쇄 소스 모델을 활용하여 오디오에서 회의록과 실행 항목을 생성하는 도구를 개발합니다.
  • 프로젝트 4: Python 코드를 최적화된 C++로 변환하여 성능을 60,000배 향상시키는 AI를 제작합니다.
  • 프로젝트 5: RAG을 활용해 회사와 관련된 모든 내용을 전문가 수준으로 이해하는 AI 지식 근로자를 개발합니다.
  • 프로젝트 6: 캡스톤 파트 A – Frontier 모델을 활용해 간단한 설명으로부터 제품 가격을 예측합니다.
  • 프로젝트 7: 캡스톤 파트 B – Frontier 모델과 경쟁할 수 있도록 미세 조정된 오픈 소스 모델을 활용해 가격 예측을 실행합니다.
  • 프로젝트 8: 캡스톤 파트 C – 여러 에이전트가 협력하며 특가 상품을 발견하고 알림을 제공하는 자율 멀티 에이전트 시스템을 구축합니다.
  • 주어진 비즈니스 문제에 대해 LLM을 선택, 학습, 적용하여 전체 솔루션을 설계하고 개발합니다.
  • RAG, 미세 조정, 에이전틱 워크플로우와 같은 최신 기술을 비교하고 대조하여 LLM 솔루션의 성능을 개선합니다.
  • 최신 Frontier 모델 10개와 오픈 소스 LLM 10개를 평가하여 주어진 작업에 가장 적합한 선택을 할 수 있도록 합니다.

Course content

8 sections • 225 lectures • 25h 16m total length

  • 1일차 - LLM 엔지니어링에 바로 뛰어들기0:37

    If you want to know:

    • How can I run large language models (LLMs) on my local computer?

    • What is Ollama and how do I set it up for running LLMs locally?

    • How can I get started with LLM engineering without complex theory?

    • What's the fastest way to begin working with open-source language models?

    • How do I set up my first local LLM environment for practical use?

    Then this lecture is for you!


    Jump straight into LLM engineering with this hands-on, practical session focused on getting you up and running with local large language models. Skip the theoretical introductions and dive directly into setting up and running open-source LLMs on your computer using Ollama. This no-nonsense approach will guide you through the essential steps of configuring your first local LLM environment, preparing you for practical AI development. Learn how to leverage tools like Langchain and Llama 2 for building functional LLM applications. Perfect for beginners eager to start their journey in AI and machine learning, this session emphasizes practical implementation over theory, getting you started with real-world LLM engineering immediately. By the end of this lecture, you'll have your own local LLM setup ready for developing chatbots and other AI applications.

  • 1일차 - Windows 및 Mac에서 로컬 LLM 배포를 위해 Ollama 셋업하기4:14

    If you want to know:

    • How do I run large language models (LLMs) locally on my computer?

    • What is Ollama and how can I use it to deploy LLMs?

    • How do I set up and install Ollama on Windows and Mac?

    • Can I run powerful language models like Llama 2 without cloud services?

    • How do I create a free AI chatbot using local LLMs?

    Then this lecture is for you!


    Learn how to deploy and run powerful large language models (LLMs) locally on your computer using Ollama, an open-source framework for local LLM deployment. This step-by-step guide covers the complete installation and setup process for both Windows and Mac systems, demonstrating how to run models like Llama 2 directly on your machine. You'll learn how to download and install Ollama, launch it through PowerShell, and create practical applications like an AI language tutor. The lecture provides hands-on experience with local LLM deployment, showing you how to leverage open-source models without relying on cloud services or paid APIs. Perfect for developers, AI enthusiasts, and anyone interested in running their own language models locally.

  • 1일차 - 로컬 LLM의 힘을 발휘하기: Ollama를 사용하여 스페인어 튜터 만들기4:07

    If you want to know:

    - How can I run powerful language models on my own computer for free?

    - What is Ollama and how can I use it to create a language learning assistant?

    - How do I set up and run different LLM models locally without cloud dependencies?

    - Which open-source language models work best for creating a language tutor?

    - How can I build a personalized language learning chatbot without coding experience?

    Then this lecture is for you!


    In this hands-on lecture, discover how to harness the power of Local Large Language Models (LLMs) using Ollama to create your own free Spanish language tutor. Learn the step-by-step process of installing and running open-source language models locally on both Mac and Windows systems. We'll explore various models including Llama 2, demonstrating how to download, install, and interact with these powerful AI tools. The lecture covers practical implementation of different language models, comparing their performance and capabilities for language teaching applications. You'll learn how to select the most suitable model for your needs, whether it's Meta's Llama 3.2, Google's Jammer, or Alibaba Cloud's Qwen. Perfect for beginners interested in AI applications, this lecture provides a foundation for building practical LLM-powered language learning tools without cloud dependencies or subscription costs.

  • 1일차 - LLM 엔지니어링 로드맵: 8주 만에 초보에서 마스터로5:45

    If you want to know:

    • How can I become a proficient LLM engineer in 8 weeks?

    • What's the step-by-step roadmap to master large language models?

    • Which tools and frameworks are essential for building LLM applications?

    • How do frontier models like GPT-4 compare to open-source alternatives?

    • What practical skills do I need to build commercial AI applications?

    Then this lecture is for you!


    This comprehensive roadmap guides you through an 8-week journey to become a skilled LLM engineer, covering both theoretical foundations and practical applications of large language models. Starting with frontier models like GPT-4 and Claude 3.5, you'll learn to build commercial AI applications using modern frameworks including Gradio, Hugging Face, and LangChain. The course progresses through essential topics such as multimodal chatbots, model selection, code generation, RAG (Retrieval Augmented Generation), and fine-tuning techniques. You'll work on real-world projects, from building AI assistants to developing autonomous agent systems that can collaborate and solve complex business problems. Each week builds upon previous knowledge, culminating in the ability to create sophisticated LLM applications using both closed-source and open-source models. The course emphasizes practical implementation, providing hands-on experience with prompt engineering, embeddings, vector databases, and transformer architectures, ensuring you gain immediately applicable skills for real-world AI development.

  • 1일차 - LLM 애플리케이션 구축: 챗봇, RAG, 에이전틱 AI 프로젝트1:49

    If you want to know:

    - How do you build practical LLM applications for real business problems?

    - What are the key components of building AI-powered chatbots and RAG systems?

    - How can you implement vector databases for efficient information retrieval?

    - How do you create agentic AI solutions that solve commercial challenges?

    - What tools and frameworks are essential for building production-ready LLM applications?

    Then this lecture is for you!


    In this hands-on lecture, you'll dive into building real-world Large Language Model (LLM) applications through practical, commercial projects. Learn to develop an intelligent airline chatbot assistant capable of ticket price lookups and multimedia interactions using prompt engineering and LangChain framework. Master the implementation of Retrieval Augmented Generation (RAG) pipelines, working with vector databases and embeddings for efficient information retrieval. Explore vector space visualizations and understand their fundamental role in modern AI applications. The lecture culminates in creating sophisticated agentic AI solutions that demonstrate practical business problem-solving capabilities. Through step-by-step guidance, you'll gain hands-on experience with Python, APIs, and essential AI engineering tools while building a GitHub portfolio of commercial-grade projects. This practical approach ensures you develop real-world skills in building LLM applications, from foundational concepts to advanced implementations in generative AI and transformer-based models.

  • 1일차 - 월스트리트에서 AI의 세계로: Ed Donner가 LLM 엔지니어가 된 여정2:07

    If you want to know:

    • How can a Wall Street veteran transition into becoming an LLM engineer?

    • What skills and experience are needed to build LLM applications?

    • How does real-world AI engineering differ from traditional software development?

    • What career paths exist in the growing field of Large Language Models?

    • Can financial sector experience translate to AI and prompt engineering?

    Then this lecture is for you!


    In this insightful introduction, Ed Donner, a seasoned tech leader with 20 years of experience in software engineering and data science, shares his journey from managing 300-person engineering teams at J.P. Morgan to becoming an accomplished LLM engineer and AI startup founder. Drawing from his extensive background spanning London, Tokyo, and New York, Ed provides valuable insights into the intersection of traditional software engineering and modern AI development. This lecture sets the foundation for an comprehensive 8-week journey into building LLM applications, covering essential aspects of prompt engineering, machine learning, and practical AI implementation. Whether you're a seasoned developer looking to transition into AI or an aspiring LLM engineer, Ed's real-world experience and successful startup exit provide a unique perspective on navigating the rapidly evolving landscape of large language models and generative AI.

  • 1일차 - 여러분의 LLM 개발 환경 설정: 툴과 모범 사례6:11

    If you want to know:

    - How do I set up a development environment for working with LLMs?

    - What tools and frameworks do I need to start building LLM applications?

    - How can I run local LLMs like Llama on my computer?

    - What are the best practices for setting up an AI development workspace?

    - How do I configure essential tools like Anaconda, Docker, and OpenAI APIs?

    Then this lecture is for you!


    This comprehensive lecture guides you through setting up a professional LLM development environment, focusing on essential tools and best practices for building AI applications. Learn how to configure a full-spec data science workspace using Anaconda or Python virtual environments, integrate crucial frameworks like LangChain, and set up local LLM implementations including Ollama. The session covers GitHub repository setup, environment configuration, OpenAI API integration, and troubleshooting strategies. You'll establish a robust development foundation for working with large language models, including both cloud-based services like ChatGPT and local open-source models. Perfect for developers looking to start building production-ready LLM applications with tools like Docker, Jupyter Lab, and vector databases. The lecture includes practical solutions for common setup challenges and ensures compatibility across different development environments.

  • 1일차 - Mac 셋업 가이드: LLM 프로젝트를 위한 Jupyter Lab과 Conda6:54

    If you want to know:

    - How do I set up a development environment for LLM projects on Mac?

    - What's the best way to configure Jupyter Lab and Conda for AI development?

    - How can I create a proper workspace for running local LLMs on MacOS?

    - How do I clone and set up an LLM engineering project repository?

    - What are the essential steps for configuring a data science environment on Mac?


    Then this lecture is for you!


    This comprehensive Mac setup guide walks you through creating a professional development environment for Large Language Model (LLM) projects. Learn how to properly configure your MacOS system with essential tools including Conda, Jupyter Lab, and Git for LLM development. The lecture covers step-by-step instructions for cloning the LLM Engineering repository, setting up Anaconda environments, and launching Jupyter Lab for interactive development. You'll master the process of creating isolated development environments using Conda, ensuring compatibility across all required packages and dependencies. Perfect for data scientists, AI developers, and anyone looking to build LLM applications on MacOS. The guide includes troubleshooting tips, best practices for environment management, and verification steps to ensure your setup is ready for LLM development workflows.

  • 1일차 - LLM 엔지니어링을 위한 Anaconda 설정: Windows 설치 가이드11:37

    If you want to know:

    - How do I set up my Windows PC for LLM development?

    - What's the best way to install Anaconda for large language model engineering?

    - How can I create a proper development environment for working with LLMs locally?

    - What are the essential steps to prepare my Windows system for LLM applications?

    - How do I configure Git and Anaconda for LLM engineering projects?

    Then this lecture is for you!


    This comprehensive Windows installation guide walks you through setting up a complete development environment for LLM engineering. Learn how to properly install Git for version control, clone the course repository, and configure Anaconda for large language model development. The lecture covers creating a dedicated conda environment with all necessary dependencies for working with local LLMs, including Python 3.11, JupyterLab, and essential AI development tools. You'll understand how to navigate the PowerShell interface, manage project directories, and verify your installation to ensure everything is properly configured for building LLM applications. Perfect for Windows users looking to establish a robust development environment for working with language models, prompt engineering, and AI model deployment.

  • 1일차 - LLM 프로젝트를 위한 대체 Python 설정: Virtualenv vs. Anaconda 가이드6:32

    If you want to know:

    - How do you set up a Python environment for LLM projects without Anaconda?

    - What's the difference between Virtualenv and Anaconda for LLM development?

    - How can you create an isolated development environment for large language models?

    - What are the steps to set up a virtual environment for both Mac and PC users?

    - How do you install and manage Python packages for LLM applications?

    Then this lecture is for you!


    Alternative Python Setup for LLM Projects: A comprehensive guide to setting up a lightweight development environment using Virtualenv as an alternative to Anaconda. This tutorial covers essential steps for both Mac and PC users, demonstrating how to create isolated Python environments for large language model applications. Learn how to initialize virtual environments, install required packages through pip, and configure JupyterLab for LLM development. The lecture includes specific command-line instructions for environment activation, package management using requirements.txt, and proper setup verification. Perfect for developers working with local LLMs, prompt engineering, and AI model deployment who prefer a simpler, more streamlined setup approach. The guide ensures compatibility with popular LLM tools, vector databases, and frameworks like Langchain while maintaining a clean, isolated development environment.

  • 1일차 - LLM 개발을 위한 OpenAI API 설정: 키, 가격 및 모범 사례7:14

    If you want to know:

    - How do you set up OpenAI API access for LLM development?

    - What's the difference between ChatGPT subscription and API pricing?

    - How much does it cost to use OpenAI's API for development?

    - What are the steps to obtain and configure OpenAI API keys?

    - How can you start building LLM applications with OpenAI's models?

    Then this lecture is for you!


    This comprehensive guide walks you through the essential process of setting up OpenAI API access for Large Language Model (LLM) development. Learn the crucial differences between ChatGPT's web interface subscription and API pricing models, understanding the cost structure for API calls and development. The lecture covers detailed steps for obtaining API keys, managing billing settings, and implementing best practices for secure key management. You'll discover how to properly configure your development environment for building LLM applications, with practical insights on API usage costs and alternatives using open-source models like Ollama. Perfect for developers looking to start building professional LLM applications with industry-leading models like GPT-4, while understanding the financial implications and security considerations of API integration.

  • 1일차 - API 키를 안전하게 저장하기 위한 .env 파일 생성5:00

    If you want to know:

    • How do you securely store API keys when building LLM applications?

    • What's the proper way to create and configure a .env file?

    • How do you set up environment variables for OpenAI API keys?

    • What are the common pitfalls when setting up API key storage on Mac and Windows?

    • How can you protect sensitive credentials in LLM development environments?

    Then this lecture is for you!


    Learn how to properly set up secure API key storage for your LLM applications through the creation and configuration of a .env file. This step-by-step guide covers both Mac and Windows environments, demonstrating essential security practices for storing OpenAI API keys and other sensitive credentials. You'll master the exact syntax requirements, understand common pitfalls, and learn platform-specific commands using tools like nano (Mac) and notepad (Windows). The lecture addresses critical security considerations for large language model development, ensuring your API keys remain protected and properly accessible in your development environment without being exposed in source control. Perfect for developers working with ChatGPT, LangChain, and other LLM frameworks who need to implement secure credential management in their AI applications.

  • 1일차 - 즉각적인 만족 프로젝트: AI 기반 웹 페이지 요약기 만들기9:31

    If you want to know:

    - How can I build my first AI-powered web application?

    - What's the easiest way to create a webpage summarizer using LLMs?

    - How do I set up a development environment for LLM applications?

    - How can I use OpenAI's API for web content summarization?

    - What tools do I need to create an AI-powered web scraper?

    Then this lecture is for you!


    In this hands-on project lecture, learn how to create an AI-powered web page summarizer using Large Language Models (LLMs). Starting with initial setup in JupyterLab and Anaconda environment configuration, you'll build a practical LLM application that scrapes and summarizes web content. The lecture covers essential development environment setup, OpenAI API integration, and implementation of BeautifulSoup for web scraping. You'll learn how to create a Website class that handles URL processing, content extraction, and text summarization using modern AI models. Perfect for developers looking to build their first practical LLM application while learning fundamental concepts in prompt engineering and AI integration. This project serves as an excellent introduction to building LLM-powered tools and working with natural language processing in a real-world context.

  • 1일차 - OpenAI의 GPT-4와 Beautiful Soup을 사용한 텍스트 요약 구현13:36

    If you want to know:

    - How can you implement text summarization using GPT-4?

    - What's the best way to combine Beautiful Soup and OpenAI for web content analysis?

    - How do system prompts and user prompts work with large language models?

    - What are the practical applications of text summarization in business?

    - How can you create an automated web content summarization system?

    Then this lecture is for you!


    Learn how to build a powerful text summarization system using OpenAI's GPT-4 and Beautiful Soup. This hands-on lecture demonstrates the implementation of document summarization using large language models (LLMs) and web scraping techniques. You'll discover how to craft effective system and user prompts, interact with OpenAI's API, and process web content using Beautiful Soup. The lecture covers practical aspects of working with LLMs, including prompt engineering, API integration, and markdown formatting for outputs. You'll explore real-world applications using popular websites and learn how to extend the solution using tools like Selenium for JavaScript-rendered pages. Perfect for developers and AI enthusiasts looking to implement practical generative AI solutions for content summarization tasks. The lecture includes code examples, best practices, and community contributions for enhanced learning.

  • 1일차 - 1일차 마무리: LLM 엔지니어링의 기억해갈 키 포인트와 다음 단계2:48

    If you want to know:

    • How do you effectively wrap up your first day of LLM engineering?

    • What are the key differences between local and cloud-based LLMs?

    • How do system prompts differ from user prompts in language models?

    • What role does text summarization play in practical LLM applications?

    • How can you transition from basic LLM concepts to advanced implementations?

    Then this lecture is for you!


    This comprehensive wrap-up session covers essential foundations in Large Language Model (LLM) engineering, from local implementation using Ollama to cloud-based solutions with OpenAI's GPT models. Learn the crucial distinction between system and user prompts while exploring practical applications in text summarization. The lecture demonstrates how to leverage both open-source and frontier models like ChatGPT, comparing their capabilities and cost implications. Discover the practical differences between running LLMs locally versus cloud deployment, understanding token usage, and implementing basic prompt engineering concepts. This session bridges fundamental concepts with advanced applications, preparing you for deeper exploration of LangChain, Hugging Face, and other essential tools in the LLM ecosystem. Perfect for developers and AI enthusiasts looking to build practical, production-ready LLM applications while understanding the trade-offs between different model deployment strategies.

  • 2일차 - LLM 엔지니어링 마스터하기: AI 개발을 위한 핵심 기술과 도구6:52

    If you want to know:

    - How do you become a proficient LLM engineer in today's AI landscape?

    - What are the essential tools and frameworks needed for LLM development?

    - How do you choose between open-source and closed-source language models?

    - What are the key techniques for implementing commercial AI solutions?

    - How can you effectively use tools like LangChain, Gradio, and Hugging Face?

    Then this lecture is for you!


    Master the fundamentals of Large Language Model (LLM) engineering in this comprehensive session focused on practical AI development skills. Learn to navigate the landscape of modern LLMs, from open-source solutions like Llama 3.1 to commercial APIs like OpenAI's ChatGPT. Discover essential frameworks including LangChain for development, Gradio for interfaces, and Hugging Face for model deployment. The lecture covers critical aspects of LLM engineering, including text summarization, fine-tuning techniques, and RAG implementations. Gain hands-on experience with Python-based AI development, understanding token management, prompt engineering, and bias mitigation. Perfect for developers with basic Python knowledge looking to build production-ready generative AI applications and chatbots. The session emphasizes practical, commercial applications while providing a solid theoretical foundation in machine learning and artificial intelligence concepts.

  • 2일차 - Frontier 모델 이해하기: GPT, Claude, 오픈 소스 LLM7:42

    If you want to know:

    - What are frontier models and how do they differ from other LLMs?

    - How do closed-source models like GPT, Claude, and Gemini compare to open-source alternatives?

    - What are the different ways to interact with and implement LLMs in your projects?

    - How do cloud APIs, managed services, and local deployment options work?

    - What role do frameworks like LangChain play in LLM development?

    Then this lecture is for you!


    Understanding Frontier Models dives deep into the landscape of modern Large Language Models (LLMs), comparing closed-source powerhouses like GPT, Claude, and Gemini with open-source alternatives such as Llama, Mixtral, and Quen. This comprehensive overview explores different implementation approaches, from cloud APIs and managed AI services to local deployment options using HuggingFace and Ollama. Learn about text summarization, fine-tuning, and practical use cases while understanding the distinctions between chat interfaces, API integrations, and framework implementations like LangChain. Perfect for developers and data scientists looking to navigate the complex ecosystem of generative AI and machine learning applications. The lecture provides hands-on insights into model selection, deployment strategies, and best practices for working with both commercial and open-source LLMs.

  • 2일차 - Ollama를 사용한 로컬 LLM 추론 방법: Jupyter를 활용한 Python 튜토리얼6:55

    If you want to know:

    - How can I run large language models locally on my computer?

    - What is Ollama and how does it compare to cloud-based LLMs?

    - How do I implement local LLM inference using Python and Jupyter?

    - Can I build a text summarization tool without using OpenAI's API?

    - How do I integrate Ollama with Python for AI applications?

    Then this lecture is for you!


    This hands-on Python tutorial demonstrates how to leverage Ollama for local Large Language Model (LLM) inference, offering a practical alternative to cloud-based solutions like ChatGPT. Learn to set up and run Llama 3.2 locally through Ollama, implement Python code for LLM interactions, and build a text summarization application without relying on OpenAI's API. The lecture covers essential concepts including API integration, local model deployment, and practical use cases for open-source LLMs. You'll explore both direct web requests and the Ollama Python package, understanding the underlying mechanics of local LLM implementation. Perfect for developers interested in generative AI applications while maintaining data privacy and reducing API costs. The tutorial includes step-by-step code examples using JupyterLab, demonstrating how to transition from cloud-based to local LLM solutions for practical machine learning applications.

  • 2일차 - 실습 LLM 작업: 텍스트 요약을 위한 OpenAI와 Ollama 비교0:36

    If you want to know:

    - How do OpenAI and Ollama compare for text summarization tasks?

    - What are the practical differences between open-source and proprietary LLMs?

    - How can you implement text summarization using different LLM frameworks?

    - What are the key considerations when choosing between different LLM APIs?

    - Which model performs better for specific summarization use cases?

    Then this lecture is for you!


    In this hands-on session, we explore practical text summarization implementations using two prominent Large Language Model (LLM) platforms: OpenAI and Ollama. Through direct comparison and real-world examples, you'll learn how to leverage both proprietary and open-source LLMs for text summarization tasks. The lecture covers essential Python implementations, API integrations, and framework-specific approaches using popular models like Llama 3.1. You'll gain practical experience with generative AI applications, understand the nuances of different LLM architectures, and learn how to evaluate model outputs effectively. This session provides valuable insights into machine learning benchmarks, model fine-tuning considerations, and best practices for implementing LLM-powered summarization solutions in production environments. Perfect for data scientists and AI practitioners looking to make informed decisions about LLM implementation choices.

  • 3일차 - Frontier AI 모델: GPT-4, Claude, Gemini, LLAMA 비교7:38

    If you want to know:

    - What are the key differences between leading AI models like GPT-4, Claude, and Gemini in 2024?

    - How do frontier AI models compare in terms of capabilities and use cases?

    - Which AI model performs best for coding, summarization, and business applications?

    - What are the strengths and limitations of open-source models like LLAMA versus proprietary models?

    - How do Claude 3 Opus and GPT-4 stack up against each other in real-world applications?

    Then this lecture is for you!


    This comprehensive lecture explores the current landscape of frontier AI models, comparing the capabilities and limitations of industry leaders including OpenAI's GPT-4, Anthropic's Claude 3 series, Google's Gemini 1.5, Meta's LLAMA, and other state-of-the-art language models. You'll learn about each model's unique strengths in areas like coding tasks, content generation, and mathematical reasoning. The lecture covers practical applications, context window sizes, and computational requirements across different models. Special attention is given to recent developments like Claude 3.5 Sonnet and its PhD-level capabilities in specific domains. You'll understand the tradeoffs between open-source and proprietary models, helping you make informed decisions for your AI implementation needs. The session includes real-world examples of model performance, hallucination risks, and practical guidelines for choosing the right model for specific use cases in business and development contexts.

  • 3일차 - 주요 LLM 비교: 강점과 비즈니스 응용점1:49

    If you want to know:

    - How do different leading LLMs like GPT-4, Claude 3, and Gemini 1.5 compare in real-world applications?

    - What are the key strengths and limitations of various AI language models?

    - How can you determine if your business problem is suitable for an LLM solution?

    - Which LLM is best suited for specific tasks like coding, summarization, or mathematical problems?

    - What factors should you consider when selecting between open-source and proprietary AI models?

    Then this lecture is for you!


    In this comprehensive comparison of leading Large Language Models (LLMs), we explore the practical applications and capabilities of frontier models including GPT-4, Claude 3, and Gemini 1.5. Through hands-on demonstrations and real-world examples, we analyze how different AI models perform across various tasks, from coding and mathematical problems to philosophical questions. The lecture provides valuable insights into model selection criteria, helping you understand the tradeoffs between open-source and proprietary solutions. We examine state-of-the-art capabilities, context windows, and computational requirements of different LLMs, enabling you to make informed decisions for your specific use cases. Special attention is given to comparing ChatGPT, Claude, Gemini, and Cohere's Command R Plus, with practical demonstrations of their strengths and limitations. This session is essential for software engineers, business leaders, and AI practitioners looking to leverage the latest developments in language models effectively.

  • 3일차 - GPT-4와 O1 Preview 비교: 성능의 주요 차이점3:54

    If you want to know:

    - What are the key performance differences between GPT-4 and GPT-4O (O1 Preview)?

    - How do frontier AI models compare in solving analytical and reasoning tasks?

    - Why do some LLMs struggle with basic counting tasks while excelling at complex reasoning?

    - What makes GPT-4O's chain-of-reasoning approach unique?

    - How are leading AI models like Claude and GPT-4 evolving in 2024?


    Then this lecture is for you!


    This comprehensive exploration delves into the performance differences between OpenAI's latest frontier models, focusing on GPT-4 and GPT-4O (formerly known as Strawberry). Through practical demonstrations and real-world examples, we examine how these large language models (LLMs) handle various tasks, from business problem analysis to precise counting and analogical reasoning. The lecture showcases GPT-4O's advanced chain-of-reasoning capabilities, demonstrating significant improvements in accuracy and problem-solving approaches compared to its predecessor. We analyze specific use cases highlighting where GPT-4O outperforms traditional models, particularly in tasks requiring detailed analysis and precise computation. This comparison provides valuable insights into the evolving landscape of AI capabilities, token processing, and the future of language models. Special attention is given to the technical aspects that differentiate these frontier models, including their handling of context, tokenization strategies, and inference methodologies.

  • 3일차 - 창의성과 코딩: GPT-4o의 캔버스 기능 활용하기6:31

    If you want to know:

    - How can GPT-4o's Canvas feature enhance your coding workflow?

    - What are the creative capabilities of modern AI models like GPT-4o and Claude?

    - How do you leverage AI assistants for interactive code development?

    - What makes GPT-4o's multimodal features stand out in practical coding scenarios?

    - How can you use AI to simplify and optimize Python code iterations?

    Then this lecture is for you!


    Dive into the creative potential of GPT-4o's Canvas feature, exploring how this frontier model transforms coding workflows and problem-solving approaches. This hands-on session demonstrates practical applications of GPT-4o's multimodal capabilities, from handling abstract concepts to generating interactive code solutions. Learn how to effectively use Canvas for collaborative coding, including Python list comprehensions, generator functions, and code optimization techniques. The lecture showcases real-time code iteration examples, comparing them with traditional approaches while highlighting GPT-4o's ability to understand context, generate example data, and propose optimized solutions. Special attention is given to the practical aspects of working with large language models (LLMs) in software development, featuring interactive demonstrations that showcase the state-of-the-art capabilities of AI in creative problem-solving and code enhancement.

  • 3일차 - Claude 3.5의 정렬 및 아티팩트 생성: 심층 분석5:26

    If you want to know:

    - How does Claude 3.5 compare to other frontier AI models like GPT-4 and Gemini?

    - What makes Claude unique in handling ethical and alignment questions?

    - How does Claude's artifact creation system work for coding tasks?

    - What are Claude's strengths and limitations in real-world applications?

    - How does Claude handle complex queries and technical challenges?


    Then this lecture is for you!


    Dive deep into Claude 3.5's capabilities and unique features in this comprehensive exploration of Anthropic's leading language model. Learn how Claude approaches complex queries, from philosophical questions to practical coding tasks, with a special focus on its distinctive alignment principles and ethical considerations. The lecture demonstrates Claude's powerful artifact creation system, showcasing real-world examples using the OpenAI API and Python programming. Compare Claude's performance against other frontier models like GPT-4 and Gemini, understanding its strengths in benchmarks and practical applications. Discover how Claude handles various challenges, from technical computations to thoughtful responses on broader socio-ethical considerations. This session provides valuable insights into state-of-the-art AI capabilities, making it essential for software engineers, data scientists, and AI enthusiasts looking to understand the current landscape of large language models and their practical applications in 2024.

  • 3일차 - AI 모델 비교: 기발하고 분석적인 작업을 위한 Gemini vs Cohere 비교4:46

    If you want to know:

    - How do Gemini and Cohere compare to other frontier AI models in 2024?

    - What are the strengths and limitations of different LLMs in handling analytical vs creative tasks?

    - How do different AI models perform in basic comprehension and counting tasks?

    - Which AI model performs best for whimsical and mathematical queries?

    - What makes certain LLMs better at specific types of tasks than others?

    Then this lecture is for you!


    This comprehensive AI model comparison lecture explores the capabilities and limitations of leading language models, focusing on Gemini and Cohere's performance in both whimsical and analytical tasks. Through practical demonstrations, we examine how these frontier models handle creative queries, mathematical problems, and basic comprehension tasks, providing direct comparisons with other state-of-the-art LLMs like GPT-4 and Claude. The lecture showcases real-world examples of each model's response patterns, highlighting their unique approaches to problem-solving and demonstrating the current state of AI capabilities in 2024. Special attention is given to analyzing response quality, context understanding, and practical use cases, offering valuable insights for software engineers, researchers, and AI enthusiasts interested in large language model benchmarking and performance optimization.

  • 3일차 - Meta AI와 Perplexity 평가: 모델 출력의 미세한 차이4:36

    If you want to know:

    - How do Meta AI and Perplexity AI compare to other frontier models like GPT-4 and Claude?

    - What are the unique strengths and limitations of Meta AI's LLAMA-based interface?

    - How well do different AI models handle basic counting and reasoning tasks?

    - What makes Perplexity different from traditional LLMs in handling real-time information?

    - Can open-source models compete with proprietary AI in image generation tasks?

    Then this lecture is for you!


    In this comprehensive evaluation of frontier AI models, we dive deep into Meta AI and Perplexity's capabilities, exploring their unique approaches to language processing and real-time information handling. The lecture demonstrates practical comparisons between these platforms and industry leaders like GPT-4 and Claude, using specific test cases including basic counting tasks and image generation prompts. We examine Meta AI's LLAMA-based implementation, showcasing its competitive image generation capabilities as an open-source alternative to proprietary models. Special attention is given to Perplexity's distinct position as a search-enhanced AI platform, highlighting its ability to process current events and provide factual, well-researched responses. Through hands-on demonstrations and comparative analysis, you'll gain practical insights into the strengths, limitations, and unique characteristics of these state-of-the-art AI models, essential knowledge for anyone working with or evaluating large language models in 2024.

  • 3일차 - LLM 리더십 챌린지: 창의적인 프롬프트를 통한 AI 모델 평가5:41

    If you want to know:

    - How do leading AI models like GPT-4, Claude 3, and Gemini 1.5 compare in real-world applications?

    - What makes certain LLMs better suited for specific tasks?

    - How are frontier models converging in capabilities and what does this mean for the future?

    - What factors beyond performance are becoming crucial in choosing between AI models?

    - How do different AI assistants handle creative leadership challenges?

    Then this lecture is for you!


    In this comprehensive exploration of modern Large Language Models (LLMs), we dive deep into comparing top AI models including GPT-4, Claude 3 Opus, and Gemini 1.5 Pro. The lecture analyzes their unique strengths, practical applications, and performance benchmarks across various tasks. Through an engaging leadership challenge experiment, we demonstrate how these frontier models approach complex, creative prompts differently. Special attention is given to emerging trends in the AI landscape, including model convergence, pricing strategies, and the growing importance of factors beyond raw performance. The session covers critical aspects of model evaluation, from token handling to context windows, providing essential insights for both technical and business audiences. Real-world examples and comparative analyses help understand how these state-of-the-art AI models are reshaping the technological landscape in 2024, with particular focus on their practical applications in business and development contexts.

  • 4일차 - 리더십 우승자 공개: 재미있는 LLM 챌린지7:50

    If you want to know:

    - Which LLM emerged as the winner in the leadership challenge between GPT-4, Claude 3 Opus, and Gemini?

    - How has the perception of AI language models evolved since the release of ChatGPT?

    - What is the significance of the "Attention is All You Need" paper in the development of modern LLMs?

    - What does "emergent intelligence" mean in the context of large language models?

    - How do frontier models like GPT-4, Claude, and Gemini actually process and generate text?

    Then this lecture is for you!


    In this comprehensive exploration of modern AI language models, we reveal the exciting results of a unique leadership challenge between frontier models GPT-4, Claude 3 Opus, and Gemini. The lecture traces the transformative journey of LLMs from the groundbreaking "Attention is All You Need" paper through the development of GPT series, ChatGPT, and contemporary multimodal models. We examine the evolution of industry perspectives on AI capabilities, from initial skepticism about "stochastic parrots" to the current understanding of emergent intelligence. The session provides detailed insights into how large language models process information, explaining core concepts like token prediction and pattern recognition that drive their impressive performance. This lecture bridges the gap between theoretical understanding and practical applications of modern AI, offering valuable perspectives for both newcomers and experienced practitioners in the field of generative AI.

  • 4일차 - AI의 여정 탐색: 초기 모델에서 Transformer까지3:02

    If you want to know:

    • How has the role of prompt engineering evolved in AI development?

    • What are the latest trends in AI collaboration and agent-based systems?

    • How do co-pilots and custom GPTs fit into the modern AI landscape?

    • What makes agentic AI different from traditional language models?

    • Why has there been a shift from individual LLMs to collaborative AI systems?

    Then this lecture is for you!


    Dive deep into the evolving landscape of artificial intelligence and large language models (LLMs) as we explore recent developments in AI collaboration and automation. This comprehensive lecture examines the transformation of prompt engineering from a highly specialized role to an accessible skill, the rise and current state of custom GPTs, and the revolutionary impact of co-pilot systems in human-AI collaboration. Special attention is given to the emerging field of agentic AI, where multiple LLMs work together with persistent memory and autonomous capabilities to solve complex problems. Learn how modern AI systems are moving beyond simple text generation to become sophisticated collaborative tools, featuring advanced natural language processing and contextual understanding. The lecture culminates with insights into practical applications of these technologies, including a preview of building multi-agent AI systems that leverage transformer models and advanced language understanding capabilities.

  • 4일차 - LLM 파라미터 이해: GPT-1에서 조단위 가중치 모델까지5:01

    If you want to know:

    • How have LLM parameters evolved from GPT-1 to modern trillion-weight models?

    • What's the significance of parameters in large language models?

    • Why do modern LLMs need billions or trillions of parameters?

    • How do parameters compare between traditional ML models and current LLMs?

    • What are the parameter counts in popular models like GPT-4, LLaMA, and Mixtral?

    Then this lecture is for you!


    Understanding Parameters in Large Language Models (LLMs) explores the fundamental building blocks that power modern artificial intelligence systems. This comprehensive lecture traces the evolutionary journey of language models from GPT-1's 117 million parameters to today's trillion-parameter frontier models. You'll learn how these parameters, or weights, function as crucial control mechanisms within LLMs, influencing their ability to understand and generate human language. The lecture compares traditional machine learning models with modern architectures, examining specific examples including GPT-2 (1.5B parameters), GPT-3 (175B parameters), GPT-4 (1.76T parameters), and open-source alternatives like LLaMA and Mixtral. Through detailed explanations of parameter scaling, you'll gain insights into why these massive neural networks require such enormous parameter counts and how they contribute to the advancement of natural language processing capabilities. This knowledge is essential for AI developers, researchers, and anyone interested in understanding the technical foundations of generative AI and transformer models.

  • 4일차 - GPT 토큰화 설명: 대형 언어 모델이 텍스트 입력을 처리하는 방식10:41

    If you want to know:

    - How do GPT and other large language models actually process text input?

    - What are tokens and why are they crucial for LLMs?

    - How does tokenization bridge the gap between human text and machine understanding?

    - What's the relationship between tokens, words, and context length?

    - How can you optimize your prompts by understanding tokenization?

    Then this lecture is for you!


    This comprehensive lecture demystifies GPT tokenization, a fundamental concept in how large language models process text. Learn how modern LLMs like GPT-4 evolved from character-based and word-based approaches to the current token-based system. Discover the practical aspects of tokenization through OpenAI's tokenizer tool, understanding how different types of text—from common words to numbers and rare terms—are processed. The lecture covers crucial concepts like context windows, token-to-word ratios, and their impact on model performance. You'll gain practical insights into token optimization, context length management, and how tokenization affects prompt engineering. Through real-world examples and demonstrations, you'll understand how tokenization influences natural language processing and machine learning capabilities, essential knowledge for anyone working with AI language models.

  • 4일차 - 컨텍스트 윈도우가 AI 언어 모델에 미치는 영향: 토큰 제한 설명3:13

    If you want to know:

    • What exactly is a context window in large language models?

    • How do token limits affect AI model performance?

    • Why can't LLMs process unlimited amounts of text?

    • How does ChatGPT maintain conversation context?

    • What's the relationship between context length and model capabilities?

    Then this lecture is for you!


    Dive deep into the crucial concept of context windows in Large Language Models (LLMs) and understand how they fundamentally shape AI performance. This comprehensive lecture explains how context length affects token processing, from basic input-output mechanisms to complex conversation handling in models like GPT-4 and ChatGPT. Learn how context windows influence natural language processing, the relationship between model parameters and token limits, and the practical implications for prompt engineering. Discover essential techniques for managing context length, optimizing prompts, and understanding how LLMs maintain conversational context through token management. Perfect for developers, AI enthusiasts, and professionals working with language models who want to maximize their understanding of these fundamental AI concepts and improve their prompt engineering skills.

  • 4일차 - AI 모델 비용 탐색: API 가격 책정 vs. 채팅 인터페이스 구독2:48

    If you want to know:

    - What's the difference between API pricing and chat interface subscriptions for AI models?

    - How do token-based costs work with GPT-4 and Claude?

    - Which pricing model is more cost-effective for different use cases?

    - How do context windows affect pricing in large language models?

    - What are the minimum API credit requirements for OpenAI and Anthropic?

    Then this lecture is for you!


    Dive deep into the economics of AI model usage, comparing subscription-based chat interfaces like ChatGPT Pro with token-based API pricing models. This comprehensive guide explores the cost structures of leading large language models including GPT-4 and Claude, breaking down how input and output tokens affect pricing. Learn about minimum credit requirements for API access, understand context window implications, and discover cost-effective strategies for both small-scale projects and larger deployments. The lecture provides practical insights into choosing between chat interfaces and APIs, helping you make informed decisions for your AI applications. Whether you're planning to use OpenAI's services, Anthropic's Claude, or exploring alternatives like Ollama, you'll gain crucial knowledge about managing AI costs effectively in 2024.

  • 4일차 - LLM 컨텍스트 윈도우 비교: GPT-4 vs Claude vs Gemini 1.5 Flash5:22

    If you want to know:

    - What are the key differences in context window sizes between GPT-4, Claude, and Gemini 1.5 Flash?

    - How do token costs compare across different LLM models?

    - What is the practical significance of a 1-million token context window?

    - How do you calculate API costs for different language models?

    - Which LLM offers the most cost-effective solution for different use cases?

    Then this lecture is for you!


    Dive deep into a comprehensive comparison of leading Large Language Models' context windows and pricing structures. This lecture analyzes the groundbreaking capabilities of Gemini 1.5 Flash with its unprecedented 1-million token context window, comparing it to Claude's 200,000 and GPT-4's 128,000 token capacities. Learn how these context windows translate to practical applications, with Gemini 1.5 Flash capable of processing nearly the complete works of Shakespeare in a single prompt. Understand the real-world cost implications of using these AI models, from Claude 3.5 Sonnet's pricing structure to GPT-4's more economical rates. The lecture breaks down token pricing, explaining how costs are calculated per million tokens for both input and output, making it essential knowledge for AI development and deployment. Discover practical insights about cost management, API usage, and selecting the right language model for specific use cases, with special attention to building scalable AI systems.

  • 4일차 - 4일차 마무리: 주요 포인트와 실용적인 인사이트 정리2:40

    If you want to know:

    - How do large language models process and understand text differently from humans?

    - What are the key limitations of current LLMs in handling basic text analysis tasks?

    - How do different AI models like GPT-4, Claude, and O1 Preview compare in their capabilities?

    - What's the relationship between tokenization and an LLM's ability to process text?

    - How do context windows affect API costs and model performance?

    Then this lecture is for you!


    This comprehensive wrap-up session explores the fundamental concepts of large language models (LLMs) and their practical applications. Learn how tokenization affects model performance, understand the crucial differences between leading frontier models like GPT-4, Claude, and O1 Preview, and master the intricacies of context windows in LLM operations. The lecture provides detailed insights into API cost considerations and demonstrates real-world applications through practical examples. Discover why certain LLMs struggle with basic text analysis tasks and how advanced models leverage chain-of-thought reasoning to overcome these limitations. Essential knowledge is shared about OpenAI and Ollama implementations, preparing you for developing commercial applications and solving complex business problems using generative AI technology. This foundation-building session bridges theoretical understanding with practical implementation, setting the stage for advanced LLM application development.

  • 5일차 - OpenAI API와 Python을 사용한 AI 기반 마케팅 브로셔 만들기3:08

    If you want to know:

    - How can I build AI-powered marketing brochures using Python and OpenAI API?

    - What is one-shot prompting and how can it improve AI content generation?

    - How do I integrate OpenAI API with Python for commercial applications?

    - How can I create automated marketing materials using large language models?

    - What are the best practices for generating professional content with AI tools?

    Then this lecture is for you!


    Learn how to build professional marketing brochures using Python and the OpenAI API in this hands-on lecture. Master practical machine learning applications by implementing one-shot prompting techniques to generate high-quality marketing content. The lecture covers essential artificial intelligence concepts, from API integration to content streaming and markdown formatting, demonstrating how to create a complete business solution. Using Jupyter notebooks, you'll develop a robust AI model that can compile information from multiple sources to create comprehensive marketing materials suitable for clients, investors, and recruitment. The session includes practical examples of data processing, API implementation, and content generation techniques using Python libraries. By the end of this lecture, you'll have built and deployed a functional AI tool that streamlines the creation of marketing materials, combining the power of large language models with practical business applications.

  • 5일차 - JupyterLab 튜토리얼: AI 기반 회사 브로셔를 위한 웹 스크래핑6:20

    If you want to know:

    - How can I use JupyterLab for web scraping in Python?

    - What's the process of building AI-powered company brochures?

    - How do you combine web scraping with large language models?

    - How can I extract and process website links using Beautiful Soup?

    - What's the best way to automate content gathering for company profiles?

    Then this lecture is for you!


    In this comprehensive JupyterLab tutorial, learn how to create AI-powered company brochures through advanced web scraping techniques using Python. The lecture demonstrates how to build a robust web scraping system that leverages Beautiful Soup and machine learning models to gather and process company information automatically. You'll work with Python libraries to extract website content, handle URL processing, and implement link parsing functionality. The tutorial showcases practical implementation using GPT-4 mini, demonstrating how to combine traditional data science approaches with modern AI tools. Through hands-on examples, you'll learn to build and deploy a system that can intelligently analyze website content, process links, and generate comprehensive company profiles. This practical session bridges the gap between basic web scraping and advanced AI-powered content generation, making it ideal for data scientists and AI practitioners looking to automate content gathering processes.

  • 5일차 - LLM의 구조화된 출력: AI 프로젝트를 위한 JSON 응답 최적화9:20

    If you want to know:

    • How do you make Large Language Models respond with structured JSON outputs?

    • What's the best way to format system prompts for JSON responses in GPT-4?

    • How can you optimize LLM responses for automated data processing?

    • What are the key differences between simple JSON requests and structured outputs in AI?

    • How do you implement one-shot prompting with JSON formatting in Python?

    Then this lecture is for you!


    This comprehensive lecture explores the implementation of structured JSON outputs in Large Language Models (LLMs), focusing on practical Python implementations with GPT-4. Learn how to create effective system prompts that generate consistent JSON responses, understand the nuances of one-shot prompting, and master the OpenAI API's response formatting capabilities. The lecture demonstrates real-world applications using Jupyter notebooks, showing how to process webpage links and transform them into structured data. You'll discover essential techniques for working with the OpenAI chat completions API, including proper message formatting and response handling. This hands-on session bridges the gap between basic LLM interactions and more sophisticated structured outputs, laying the groundwork for building advanced AI agents and automated data processing systems. Perfect for data scientists and machine learning practitioners looking to enhance their AI development skills with practical, production-ready techniques.

  • 5일차 - 브로셔 콘텐츠에 대한 응답 생성 및 포맷팅8:39

    If you want to know:

    - How to create AI-powered content generation systems using Python?

    - How to integrate Large Language Models for automated brochure creation?

    - How to build a system that analyzes websites and generates marketing materials?

    - How to combine multiple AI calls to create more sophisticated applications?

    - How to use Jupyter notebooks for developing AI content generation tools?

    Then this lecture is for you!


    Learn how to develop an advanced content generation system using Python and Large Language Models. This hands-on session demonstrates how to create a sophisticated brochure generation tool that leverages machine learning and artificial intelligence. You'll master the process of building functions that analyze website content, extract relevant information using AI models, and automatically generate professional marketing materials. The lecture covers implementing multiple API calls to AI models, handling website data processing, and creating formatted responses using Jupyter notebooks. Through practical examples, you'll understand how to combine different AI tools and Python libraries to build and deploy an intelligent content creation system. Perfect for data scientists and developers looking to create practical AI applications for business use cases.

  • 5일차 - 최종 조정: JupyterLab에서 Markdown 최적화 및 스트리밍9:50

    If you want to know:

    - How do you implement streaming responses in JupyterLab with LLMs?

    - What's the best way to optimize Markdown display for streaming AI responses?

    - How can you create dynamic, real-time AI responses in Jupyter notebooks?

    - How do you modify system prompts to control AI output tone and style?

    - What are the practical applications of multi-step LLM processes in business?

    Then this lecture is for you!


    Master advanced JupyterLab techniques for optimizing Large Language Model interactions through streaming responses and Markdown enhancements. Learn how to implement OpenAI's streaming functionality using Python, enabling real-time, typewriter-style outputs in your Jupyter notebooks. This hands-on session covers essential machine learning workflows, including system prompt engineering for controlling AI output tone, multi-step LLM processes, and practical business applications. Discover how to build sophisticated AI workflows by combining multiple LLM calls, data synthesis, and content generation. Perfect for data scientists and AI developers looking to enhance their machine learning projects with advanced Jupyter implementations and transformer model integrations. The lecture demonstrates real-world applications using popular AI tools and frameworks, including OpenAI's API, Claude, and Hugging Face, while emphasizing practical deployment strategies for AI model training and development.

  • 5일차 - 멀티샷 프롬프트 마스터하기: AI 프로젝트에서 LLM 신뢰성 향상4:22

    If you want to know:

    - How can multi-shot prompting improve LLM reliability?

    - What's the difference between one-shot and multi-shot prompting in AI applications?

    - How to enhance prompt engineering techniques for better AI responses?

    - What are the best practices for implementing multi-shot prompting in generative AI projects?

    - How can you optimize LLM outputs through advanced prompting strategies?

    Then this lecture is for you!


    Master the art of multi-shot prompting to significantly enhance Large Language Model (LLM) reliability in your AI projects. This comprehensive session explores advanced prompt engineering techniques, focusing on implementing multiple examples in your prompts for improved AI response accuracy. Learn how to transition from basic one-shot prompting to more sophisticated multi-shot approaches, understanding their impact on natural language processing outcomes. The lecture covers practical use cases, demonstrating how multi-shot prompting strengthens an LLM's ability to generate more consistent and reliable outputs. Discover best practices for structured outputs, iterative prompt development, and the strategic implementation of system prompts across different AI applications. Special attention is given to real-world applications, including brochure generation and language translation scenarios, providing hands-on experience with foundation models and open-source LLMs. This session equips prompt engineers with essential engineering skills for mastering generative AI implementations and optimizing AI responses through advanced prompting strategies.

  • 5일차 - 과제: 여러분만의 맞춤형 LLM 기반 튜터 개발하기4:07

    If you want to know:

    - How can I create my own personalized AI tutor using LLMs?

    - What's the difference between using GPT and open-source LLAMA for custom tutoring?

    - How do I implement streaming responses with Markdown formatting in JupyterLab?

    - How can I build an interactive tool for technical and data science learning?

    - What are the best practices for developing a customized LLM-based learning assistant?

    Then this lecture is for you!


    Learn how to develop your own personalized LLM-based tutor in this hands-on assignment focused on practical prompt engineering and AI implementation. Using both GPT and open-source LLAMA models, you'll create a custom learning assistant that can answer questions about code, LLMs, and technical concepts. The lecture guides you through setting up the environment in JupyterLab, implementing streaming responses with Markdown formatting, and comparing outputs between different language models. You'll learn essential prompt engineering techniques while building a practical tool that serves as your personal technical co-pilot. The assignment includes working with foundation models, natural language processing, and iterative prompt development, providing you with real-world experience in creating AI-powered educational tools. Perfect for those looking to master prompt engineering while developing practical AI applications.

  • 5일차 - 1주차 마무리: 성취 및 다음 단계2:56

    If you want to know:

    • How do Large Language Models (LLMs) handle different types of prompts?

    • What are the key differences between single-shot and multi-shot prompting?

    • How can you effectively use system prompts to control AI responses?

    • What are the practical applications of OpenAI and Ollama APIs?

    • How does tokenization impact LLM performance?

    Then this lecture is for you!


    This comprehensive wrap-up lecture consolidates the fundamental concepts of Large Language Models (LLMs) and prompt engineering covered in the first week. Students will review crucial aspects of transformer architecture, tokenization principles, and context window optimization. The lecture covers practical implementations using OpenAI's API, including advanced features like streaming and markdown integration. Participants will understand the strategic use of system prompts for tone control and instruction setting, along with the differences between single-shot and multi-shot prompting techniques. The session also explores Ollama API implementation for local model deployment, preparing learners for advanced topics in retrieval augmented generation and generative AI applications. The lecture concludes with a preview of upcoming content, including multi-modal customer support agents and data science UI development using Gradio. This session bridges foundational knowledge with practical applications in natural language processing and artificial intelligence.

Requirements

  • 이 강의는 Python으로 진행됩니다. Python의 기초를 다루지 않으므로, Python에 대한 기본 지식이 필요합니다.

Description

[꼭 읽어주세요] 한글 AI 자막 강의란?

  • 유데미의 한국어 [자동] AI 자막 서비스로 제공되는 강의입니다.

  • 강의에 대한 질문사항은 강사님이 확인하실 수 있도록 Q&A 게시판에 영어로 남겨주시기 바랍니다.


생성형 AI와 LLM 마스터하기: 8주간의 실습 여정


AI 실무 프로젝트를 통해 커리어를 발전시키고,
이 분야의 베테랑인 Ed Donner 강사님이 이끄는 강의를 통해 생성형 AI와 최첨단 기술을 마스터합니다.
20개 이상의 혁신적인 모델을 실험하며, RAG, QLoRA, 에이전트와 같은 최신 기술을 익혀보세요!


1. 무엇을 배우나요?


  • 최첨단 모델과 프레임워크를 사용해 고급 생성형 AI 제품을 개발합니다.

  • Frontier 및 오픈 소스 모델을 포함한 20개 이상의 혁신적인 AI 모델을 실험합니다.

  • HuggingFace, LangChain, Gradio와 같은 플랫폼을 능숙하게 활용합니다.

  • RAG(검색 기반 생성), QLoRA 미세 조정, 에이전트와 같은 최신 기술을 구현합니다.

  • 실무 기반의 AI 애플리케이션을 제작합니다:

  • 텍스트, 음성, 이미지와 상호작용하는 멀티모달 고객 지원 에이전트

  • 공유 드라이브 데이터를 기반으로 기업 질문에 답할 수 있는 AI 지식 근로자

  • 소프트웨어를 최적화해 성능을 60,000배 개선하는 AI 프로그래머

  • 보지 못한 제품의 가격을 정확히 예측하는 이커머스 애플리케이션

  • 추론에서 학습으로 전환, Frontier 및 오픈 소스 모델을 모두 미세 조정합니다.

  • UI와 고급 기능을 갖춘 AI 제품을 프로덕션에 배포합니다.

  • AI와 LLM 엔지니어링 역량을 강화해 업계의 최전선에 자리합니다.


2. 프로젝트 소개:


  1. 프로젝트 1: 기업 웹사이트를 지능적으로 스크래핑하고 탐색하는 AI 기반 브로셔 생성기.

  2. 프로젝트 2: UI와 기능 호출을 사용하는 항공사 멀티모달 고객 지원 에이전트.

  3. 프로젝트 3: 오디오에서 회의록과 실행 항목을 생성하는 오픈 소스 및 폐쇄 소스 모델 기반 도구.

  4. 프로젝트 4: Python 코드를 최적화된 C++로 변환하여 성능을 60,000배 향상시키는 AI.

  5. 프로젝트 5: RAG를 사용하여 회사 관련 모든 정보에 대한 전문가가 되는 AI 지식 근로자.

  6. 프로젝트 6: 캡스톤 파트 A – Frontier 모델을 사용하여 간단한 설명으로부터 제품 가격 예측.

  7. 프로젝트 7: 캡스톤 파트 B – Frontier와 가격 예측에서 경쟁하기 위한 미세 조정된 오픈 소스 모델.

  8. 프로젝트 8: 캡스톤 파트 C – 모델과 협력하여 특가 상품을 발견하고 알림을 제공하는 자율 에이전트 시스템.


3. 왜 이 강의인가요?


  1. 실습 중심 학습: 실무에서 사용할 수 있는 AI 애플리케이션을 직접 구축하며 배우는 가장 효과적인 학습 방식.

  2. 최신 기술 습득: RAG, QLoRA, 에이전트와 같은 최신 프레임워크와 기술을 선도적으로 익힙니다.

  3. 접근성 높은 콘텐츠: 모든 수준의 학습자를 위해 설계되었습니다. 단계별 안내, 실습 과제, 치트시트, 다양한 리소스를 제공합니다.

  4. 고급 수학 불필요: 실질적인 응용에 초점을 맞춰 미적분이나 선형대수 지식 없이도 LLM 엔지니어링을 마스터할 수 있습니다.


4. 강사 소개

안녕하세요, 저는 Ed Donner 입니다. 20년 이상의 경력을 가진 AI 및 기술 분야의 기업가이자 리더입니다.
AI 스타트업을 설립하고 성공적으로 매각했으며, 또 다른 스타트업을 창업하여 전 세계 주요 금융기관 및 스타트업에서 팀을 이끌어 왔습니다.
이 흥미로운 분야로 더 많은 사람을 이끌고, 업계의 선두주자가 될 수 있도록 돕는 것에 열정을 가지고 있습니다.


5. 강의 커리큘럼


1주차: 기초 및 첫 번째 프로젝트

  • Transformer의 기본 개념을 학습합니다.

  • 주요 Frontier 모델 6개를 실험합니다.

  • 웹을 스크래핑하고 판매 브로셔를 생성하는 비즈니스 AI 제품을 만듭니다.


2주차: Frontier API와 고객 서비스 챗봇

  • Frontier API를 탐색하고 3가지 주요 모델과 상호작용합니다.

  • 텍스트, 이미지, 오디오와 상호작용하며 툴이나 에이전트를 활용하는 챗봇을 개발합니다.


3주차: 오픈 소스 모델 활용

  • HuggingFace를 통해 오픈 소스 모델을 탐구합니다.

  • 번역부터 이미지 생성까지 10가지 생성형 AI 사용 사례를 해결합니다.

  • 회의록과 액션 아이템을 생성하는 제품을 구축합니다.


4주차: LLM 선택과 코드 생성

  • LLM간의 차이점을 이해하고, 주어진 비즈니스 작업에 가장 적합한 LLM을 선택하는 방법을 배웁니다.

  • LLM을 사용하여 코드를 생성하고, Python 코드를 C++로 변환하는 제품을 구축하여 성능을 60,000배 이상 향상시킵니다.


5주차: RAG (검색 기반 생성)

  • RAG을 마스터하여 여러분의 솔루션의 정확도를 개선합니다.

  • 벡터 임베딩에 능숙해지고, 인기 있는 오픈 소스 벡터 데이터스토어에서 벡터를 탐색합니다.

  • 시장의 실제 제품과 유사한 풀 비즈니스 솔루션을 구축합니다.


6주차: 트레이닝으로 전환

  • 추론에서 트레이닝으로 전환합니다.

  • Frontier 모델을 미세 조정해 실제 비즈니스 문제를 해결합니다.

  • 자신만의 특화된 모델을 구축하여 여러분의 AI 여정에서 중요한 이정표를 달성합니다.


7주차: 고급 트레이닝 기술

  • QLoRA 미세 조정과 같은 고급 학습 기술을 배웁니다.

  • 특정 작업에서 Frontier 모델을 능가하는 오픈 소스 모델을 학습합니다.

  • 기술을 한 단계 더 발전시키는 도전적인 프로젝트를 해결합니다.


8주차: 배포 및 최종화

  • UI가 완성된 상업용 제품을 프로덕션에 배포합니다.

  • 에이전트를 활용해 기능을 확장합니다.

  • 첫 번째 프로덕션화된, 에이전트화된, 미세 조정된 LLM 모델을 배포합니다.

  • AI와 LLM 엔지니어링의 마스터한 것을 기념하고, 여러분의 커리어의 다음 단계에 대비합니다.

Who this course is for:

  • 생성형 AI와 LLM 분야에 진입하고자 하는 열정적인 AI 엔지니어 및 데이터 사이언티스트 지망생들
  • 빠르게 변화하는 AI 환경에서 경쟁력을 유지하고자 하는 전문가들
  • 실용적이고 실습 중심의 경험을 통해 고급 AI 애플리케이션을 개발하고자 하는 개발자들

Report abuse