Generative AI: Beginner to Advanced with LLMs, RAG & Agents

Giới thiệu khóa học

Generative AI is changing how software is built, how information is accessed, and how businesses solve everyday problems. But building a useful AI application involves much more than simply connecting an LLM to a chatbot. Generative AI: Beginner to Advanced with LLMs, RAG & AI Agents takes you through the complete journey of building modern AI-powered applications, from understanding the foundations of Large Language Models to deploying production-ready GenAI systems. You’ll begin by understanding how modern LLMs work, including Transformers, tokens, embeddings, attention, and the LLM training pipeline. You’ll also learn the difference between pretraining, fine-tuning, and instruction tuning. Next, you’ll learn how to communicate effectively with LLMs through Prompt Engineering. You’ll explore zero-shot and few-shot prompting, role prompting, prompt templates, prompt chaining, structured outputs, and tool calling. You’ll then put these concepts into practice by building applications such as an AI summarizer, coding assistant, and research assistant. The course then moves into application development. You’ll learn how LLM APIs, REST APIs, JSON, client-server architecture, and FastAPI work together to turn AI models into usable applications. From there, you’ll explore popular frameworks such as LangChain and LlamaIndex and build projects including a Chat with PDF application, a document summarizer, and an AI knowledge assistant. A major part of the course focuses on Retrieval-Augmented Generation (RAG). You’ll understand the complete RAG pipeline—from document processing and chunking to embeddings, vector databases, retrieval, and LLM generation. You’ll also explore advanced techniques such as hybrid search, reranking, parent-child chunking, multi-query retrieval, and context compression. You’ll then move beyond traditional RAG into Graph RAG, learning how knowledge graphs can improve information retrieval and building a graph-based document question-answering system. The course also introduces LLM fine-tuning with LoRA and QLoRA, giving you hands-on experience with parameter-efficient fine-tuning using Hugging Face Transformers and PEFT. As your systems become more advanced, you’ll learn how to evaluate and secure AI applications using tools such as LangSmith and Ragas. Topics include faithfulness, relevance, hallucination detection, answer correctness, prompt injection defense, output validation, and guardrails. You’ll then explore AI agents and autonomous systems, learning how agents can plan tasks, use tools, and execute multi-step workflows. You’ll build projects such as an AI research agent, travel planning agent, and multi-tool assistant using modern agent frameworks.

Bạn sẽ học được gì