🚀 Full Stack Generative & Agentic AI with Python

A hands-on learning repository exploring modern AI patterns including LLM integration, RAG systems, LangGraph workflows, and chain-of-thought reasoning.

---

📋 Table of Contents

• Overview
• Prerequisites
• Getting Started
• Project Structure
• Learning Modules
• Tech Stack
• Resources

---

Overview

This repository contains my learning experiments and code samples from exploring generative AI and agentic patterns. It covers a wide range of topics from basic LLM integration to advanced concepts like RAG (Retrieval Augmented Generation) and LangGraph workflows.

⚠️ Note: This is a learning repository. Code may be experimental or incomplete as it reflects my learning journey.

---

Prerequisites

Before running the examples, ensure you have the following installed:

Required

Tool	Version	Purpose
Python	3.10+	Core runtime for Python examples
Node.js	18+	Runtime for JavaScript examples
Ollama	Latest	Local LLM inference server

Optional (for specific modules)

Tool	Purpose
Docker	Running Qdrant vector database
Qdrant	Vector database for RAG examples
Gemini API Key	Google Gemini examples

Recommended Ollama Models

# Chat models
ollama pull qwen2.5:14b
ollama pull gemma3:27b
ollama pull qwen3:30b

# Vision models
ollama pull qwen3-vl:2b

# Embedding models
ollama pull bge-m3

# Coding models
ollama pull qwen3-coder

---

Getting Started

1. Clone the Repository

git clone https://github.com/sangleshubham/Learn-generative-Agentic-AI.git
cd Learn-generative-Agentic-AI

2. Set Up Environment Variables

cp .env.example .env
# Edit .env with your API keys

3. Install Python Dependencies

python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate
pip install -r requirements.txt

4. Start Ollama

ollama serve

5. Start Qdrant (for RAG examples)

docker compose -f 11-rag-qdrant-python/docker-compose.yml up -d

---

Project Structure

agentic-generative-ai-course/
├── 01-ollama-fastapi/           # FastAPI + Ollama integration
├── 02-hugging-face-transformers/# HuggingFace Transformers pipeline
├── 03-chain-of-thought-agent/   # CoT agent with tool calling
├── 04-ollama-nodejs-agent/      # Node.js agent implementation
├── 05-multi-modal-vision/       # Vision/image understanding
├── 06-langgraph-nodejs/         # LangGraph with Node.js
├── 07-langgraph-python/         # LangGraph with Python
├── 08-gemini-few-shot-prompting/# Gemini + few-shot prompting
├── 09-chain-of-thought/         # Basic CoT prompting
├── 10-token-encoding/           # Token encoding/decoding
├── 11-rag-qdrant-python/        # RAG with Qdrant (Python)
├── 12-rag-qdrant-nodejs/        # RAG with Qdrant (Node.js)
├── 13-RAG-project-file-search-engine/# RAG File Search Engine
├── .env.example                 # Environment configuration template
├── requirements.txt             # Python dependencies
└── README.md                    # This file

---

Learning Modules

🔹 01 - Ollama + FastAPI

Basic LLM chat API using FastAPI and Ollama client.

• Tech: Python, FastAPI, Ollama
• Concepts: REST API, chat completions

🔹 02 - HuggingFace Transformers

Using HuggingFace pipelines for multimodal inference.

• Tech: Python, Transformers, Gemma-3
• Concepts: Image-text-to-text pipeline

🔹 03 - Chain-of-Thought Agent

AI agent using CoT prompting with tool calling capabilities.

• Tech: Python, OpenAI-compatible API
• Concepts: ReAct pattern, function calling, structured output

🔹 04 - Ollama Node.js Agent

Interactive agent implementation in Node.js with Zod schema validation.

• Tech: Node.js, OpenAI SDK, Zod
• Concepts: Schema validation, system command execution

🔹 05 - Multi-Modal Vision

Image understanding using vision language models.

• Tech: Node.js, OpenAI SDK
• Concepts: Base64 image encoding, vision models

🔹 06 - LangGraph (Node.js)

LangGraph workflow exploration in Node.js.

• Tech: Node.js, LangGraph
• Concepts: State graphs, graph workflows

🔹 07 - LangGraph (Python)

State machines and conditional routing with LangGraph.

• Tech: Python, LangGraph, LangChain
• Concepts: StateGraph, conditional edges, node chaining

🔹 08 - Gemini Few-Shot Prompting

Using Gemini API with OpenAI SDK and few-shot examples.

• Tech: Python, Gemini API
• Concepts: Few-shot learning, structured prompts

🔹 09 - Chain-of-Thought

Basic chain-of-thought reasoning patterns.

• Tech: Python, Ollama/Gemini
• Concepts: CoT prompting, JSON structured output

🔹 10 - Token Encoding

Understanding tokenization using tiktoken.

• Tech: Python, tiktoken
• Concepts: Token encoding/decoding, token counting

🔹 11 - RAG with Qdrant (Python)

Complete RAG pipeline using Qdrant vector database.

• Tech: Python, LangChain, Qdrant, Ollama
• Concepts: PDF loading, text splitting, embeddings, similarity search

🔹 12 - RAG with Qdrant (Node.js)

RAG implementation in Node.js with EPUB/PDF support.

• Tech: Node.js, LangChain.js, Qdrant
• Concepts: Document loaders, vector stores, RAG queries

🔹 13 - RAG Project File Search Engine

RAG-based search engine for querying file contents using local LLMs.

• Tech: Node.js, LangChain, Qdrant, Ollama
• Concepts: RAG, Vector Search, File Ingestion, CLI Interface

---

Tech Stack

Category	Technologies
Languages	Python 3.10+, Node.js 18+
LLM Providers	Ollama (local), Google Gemini
Frameworks	FastAPI, LangChain, LangGraph
Vector Database	Qdrant
Embeddings	BGE-M3 (via Ollama)
Validation	Pydantic, Zod

---

Resources

• 📚 Udemy Course
• 🦙 Ollama Documentation
• 🔗 LangChain Documentation
• 📊 LangGraph Documentation
• 🗄️ Qdrant Documentation

---

_{Made with ❤️ while learning AI}