AI Learning Projects (RAG + LLM Apps)

This repository contains a collection of small AI projects built while learning Retrieval-Augmented Generation (RAG), Large Language Models (LLMs), and AI application development in Python.

The projects include both API-based systems and interactive AI applications, developed as part of my learning journey in modern AI engineering.

Some projects were developed as part of the IBM “Build RAG Applications: Get Started” learning path.

---

Projects Included

1. Build a RAG API with FastAPI

A local AI pipeline that implements the full Retrieval-Augmented Generation (RAG) workflow.

What it does:

• Loads and stores documents locally
• Retrieves relevant information from documents
• Uses an LLM to generate final answers
• Runs as a REST API using FastAPI

Key idea:

Build AI systems that answer questions from:

• internal documents
• support tickets
• product catalogs

---

2. AI Image & Vision

• Image Captioning using BLIP model
• Image Classification using ResNet18
• Simple AI inference pipelines

---

3. RAG Question Answering Bot

• PDF upload and processing
• Text chunking and embeddings
• Vector database (ChromaDB)
• AI-powered question answering
• Built with LangChain + Gradio

---

4. AI Icebreaker Bot

A networking assistant that generates personalized conversation starters from profile data.

• Mock LinkedIn-style data
• RAG-based retrieval system
• Ollama LLM integration
• Gradio UI

---

Technologies Used

• Python
• FastAPI
• Gradio
• LangChain
• LlamaIndex
• Hugging Face Transformers
• Ollama (Llama3 / TinyLlama)
• ChromaDB
• Sentence Transformers

---

What I Learned

• How Retrieval-Augmented Generation (RAG) works end-to-end
• Difference between LangChain and LlamaIndex
• How embeddings and vector databases work
• Building REST APIs for AI systems
• Running local LLMs using Ollama
• Creating interactive AI apps using Gradio

---

Framework Understanding

• LangChain → Low-level control, full RAG pipeline building
• LlamaIndex → High-level automated RAG framework
• Hugging Face → Pre-trained models and embeddings hub

---

Goal of This Repository

To build practical experience in:

• AI system design
• RAG pipelines
• Local LLM applications
• End-to-end AI app development

This repository serves as a learning and experimentation space for AI engineering concepts.

---

Note

All projects are for learning purposes and are continuously improved as I explore more advanced AI concepts.