Tutor-AI

Tutor-AI is a study companion app that demonstrates the power of Retrieval Augmented Generation (RAG) for helping students understand and explore complex material. By combining cutting-edge AI with your own source documents, Tutor-AI answers questions contextually and accurately — making studying smarter and more efficient.

🔍 How It Works

Tutor-AI uses:

• Google Gemini for powerful language understanding
• LangGraph to orchestrate multi-step reasoning over documents
• React + Shadcn to deliver a clean, modern user interface

This project showcases the efficacy of RAG-based systems in educational settings, turning raw content into interactive, intelligent study help.

Created by Pavundeep Dhillon.

1. Usage

To use the final solution, clone the reppository and run the tutor-ai folder for the React application and the chat_api.py file for the API.

2. Specifications

2.1 Technolgies/Libraries Used

• Python
  • LangChain
    • LangSmith
    • LangGraph
  • FastAPI
  • Uvicorn
  • Pydantic
  • Boto3
  • pip & miniconda
• Javascript/CSS/HTML
  • React
  • Shadcn
• Cloud Services
  • AWS
    • S3

2.2 File Structure

study-buddy/
├── tutorai/
│ ├── app/
│ │  ├── favicon.ico
│ │  ├── globals.css
│ │  ├── layout.tsx
│ │  └── page.tsx
│ ├── components/
│ │  ├── ui/
│ │  ├── AppSidebar.tsx
│ │  ├── ChatWindow.tsx
│ │  ├── Message.tsx
│ │  └── UploadButton.tsx
│ ├── hooks/
│ │  ├── use-mobile.tsx
│ │  └── use-toast.ts
│ ├── interfaces/
│ │  └── IMessage.ts
│ ├── lib/
│ │  └── utils.ts
├── api.py
├── brain.py
├── chat_api.py
├── chat_service.py
├── cleaner.py
├── package-lock.json
├── package.json
├── README.md
├── state.py
├── storage.py
└── tests.ipynb 

3. Process

3.1 Problem Definition

The specific student needs or study challenges the app is designed to address, and why RAG is a good fit.

3.2 Data Preparation & Chunking Strategy

How study material is sourced, cleaned, chunked, and embedded — crucial for RAG performance

3.3 LangGraph and Prompt Engineering

How LangGraph was used to build reasoning workflows and how prompts were iterated for better responses.

3.4 Frontend Design

Discuss integrating the AI backend with the React + Shadcn frontend and designing a smooth, helpful UI.

3.5 Testing, Evaluation & Iteration

How I tested retrieval quality, response relevance, and overall user experience

4. Improvements and lessons learned