Gen AI: RAG, Agents and MCP

This course covers how to use Generative AI with prompt engineering, how to build RAG (Retrieval Augmented Generation) applications, how to build AI Agents, how to use MCP (Model Context Protocol), and how to use AI pair programmer (like GitHub Copilot).

You learn how to apply all these concepts in a Capstone project.

At the end of this course, you will have the skills to build AI applications like RAG systems and Agents.

Course Details

Prerequisite	Basic knowledge of Python
Course Fee	Rs. 9,500/- (Includes digital course material)
Digital or Physical Certificate Fee	Rs. 300/-
Source code of examples	All examples demonstrated in the classroom are made available through GitHub repository

Detailed Syllabus

Foundations of Generative AI

What is AI - History and important milestones
Machine Learning, Deep Learning and Generative AI
Introduction to Generative AI, Foundation Models, Large Language Models, Image Models and Code Models
Using LLMs like GPT, Gemini and their characteristics
Basics of Transformer architecture
Understanding key terms - Tokens, Parameters, Embeddings, Context window etc.
Capabilities of LLMs
Limitations of LLMs
Leaderboards, Regulations and Future trends

Prompt Engineering

What is Prompt and Prompt Engineering
Components of a Prompt
Getting structured output
Prompting Patterns - Zero-shot, Few-shot, Chain-of-thoughts, Persona, Recipe, Reflection Meta Language Creation, Flipped Interaction, Refusal Breaker, Game Play and more
Prompt Optimization

Accessing LLMs using Python

Accessing GPT using Open AI API (Proprietary API)
Accessing Gemini using Gemini API
Understanding HuggingFace Hub and its models
What is LangChain and how to access GPT and Gemini using it
Accessing Open Source Models using LangChain
How to download models from HuggingFace to local system and access them - transformers and pipeline()
Installing ollama tool and downloading models using it
Accessing models through ollama
Understanding LLM Parameters - temperature etc.
How to use LLM parameters with models
How to build chatbot to have conversation with LLM
Working with streamlit to build web apps that access AI Models

Retrieval Augmented Generation (RAG)

What is Retrieval Augmented Generation(RAG) and its importance
Different components related to RAG - Loaders, Splitters, Embeddings, Vector Database, Similarity Search etc.
Implementing RAG using text file, pdfs, webpages and database
Building a chatbot using RAG and PDF
Introduction to Fine-Tuning
Introduction to RLHF

AI Pair Programming

What is AI Pair Programming
How to use Github Copilot or Gemini Code Assist with Python and other languages
How to integrate AI Pair Programmer with IDEs
How to use Chat feature to generate code
How to enhance code generated by AI Pair Programmer
How to use participants and slash commands
How to generate documentation using Copilot
How to generate test cases
How to fix errors

Building AI Agents

What is AI Agent
Understanding agent concepts - Reasoning, Action and Planning
How to make LLMs use tools - tool calling LLMs
Using LangGraph to build nodes and edges
Multi-step workflows with LangGraph
Building AI Agents that access tools to take actions autonomously
Working with Memory - MemorySaver
Getting results in a structured format using Pydantic

MCP (Model Context Protocol)

What is MCP and its role in AI ecosystem
MCP SDKs and Development tools
What is MCP server and client
Understanding STDIO and Streamable HTTP
How to build MCP server that use custom functions
How to access MCP Server using MCP Client

Capstone Project

Understand the requirements and required technologies to be used
Understand the data ingestion process
How to use Embeddings Model, create vector db and persist it
Develop retrieval pipeline
Select model and API to be used
Build RAG application that answers questions based on the context
Provide Web UI using Streamlit