Overview

Welcome to the Gen AI section of CS Notes. This part of the notes is built to take you from knowing nothing about Generative AI to being able to build your own AI agents, tools, and projects, step by step, in simple language.

You don’t need any prior AI experience to start here. We begin with the basics and slowly move toward more advanced and practical topics. Below is a simple breakdown of everything this section covers, so you know what to expect and where you are headed.

What You Will Learn

1. Core Foundations of Generative AI: We start with the basic building blocks behind Generative AI, such as what a token is, how a model turns words into numbers, and how AI models actually generate text.

2. API Setup & Integration: You will learn how to get API access from AI providers and connect your code to them, so you can start sending requests and getting AI generated responses.

3. Advanced Prompt Engineering Techniques: This covers different ways of writing prompts, so you can guide an AI model to give you better, more accurate, and more useful answers.

4. Prompt Serialization & Instruction Formats: Here you will learn how prompts are structured and formatted behind the scenes, so different AI models can understand and follow your instructions properly.

5. Local LLM Deployment & API Integration: You will learn how to run AI models directly on your own computer instead of always depending on an online service.

6. Running LLMs via Hugging Face Hub: This covers how to use Hugging Face’s tools and model library to download and run AI models yourself.

7. Building AI Agents and Agentic Workflows: Here we go from a simple AI model to a full AI Agent, one that can use tools, make decisions, and complete tasks on its own.

8. Building Chat with PDF Project using RAG (Retrieval-Augmented Generation): A hands on project where you build a chatbot that can read a PDF file and answer questions about it.

9. Scalable RAG with Async Queues & Distributed Workers: This shows how to make a RAG based system strong enough to handle many users and large amounts of data at the same time.

10. Multi Modal Agents: You will learn how to build agents that understand more than just plain text, including things like images and other types of data.

11. Building Agentic Workflows with LangGraph: Here you will use a tool called LangGraph to build agents that follow a clear, organized, step-by-step workflow.

12. Checkpointing Workflows in LangGraph with MongoDB: This covers how to save an agent’s progress and pick up right where it left off, using a database called MongoDB.

13. The Memory Layer - Building Short, Long, and Semantic Memory in AI Agents: You will learn how to give an AI agent different kinds of memory, so it can remember things during a conversation and even after it ends.

14. Graph Memory and Knowledge Graphs In AI Agents: Here we look at storing an agent’s knowledge as a connected graph, so it can understand how different pieces of information relate to each other.

15. Conversational Agentic AI with Voice Agents and Chained Patterns: This covers building agents that can talk using voice and follow a chain of connected steps to complete a task.

16. Model Context Protocol (MCP): A standard way for AI tools and agents to share context and connect with each other smoothly.

17. Agent SDK: A ready made toolkit that makes it easier and faster to build your own AI agents.

By the end of this section, you will not only understand how Generative AI works under the hood, but you will also be able to build your own AI powered tools and agents from scratch. Let’s get started with the foundations.