AI Lesson Note Generator (RAG)

A full-stack web application that allows educators to generate comprehensive lesson notes from various source materials (PDFs, DOCX, audio, images) using a Retrieval-Augmented Generation (RAG) pipeline.

This application provides a single-page app (SPA) experience for creating lessons, uploading source materials, ingesting them into a vector database, and generating detailed, editable lesson plans using large language models.

Key Features

• Lesson Management: Create, list, update, and delete lessons.
• Multi-format Source Upload: Upload sources in various formats including PDF, DOCX, MP3, WAV, and plain text.
• AJAX Uploads with Progress: Files are uploaded asynchronously with a visual progress bar.
• Per-Lesson Data Silos: Each lesson has its own dedicated vector collection (using ChromaDB), ensuring content remains isolated.
• RAG-based Content Generation:
  • Targeted Generation: Use a simple prompt to generate or refine specific parts of a lesson note.
  • Full Lesson Generation: A dedicated button synthesizes all ingested content for a lesson into a complete, well-structured lesson plan based on a comprehensive prompt.
• Markdown Editor: Lesson notes are rendered from Markdown in a "view mode" and can be edited in a raw Markdown "edit mode".
• API-driven Backend: A robust Flask backend provides a full suite of API endpoints for all frontend functionality.

How It Works: The RAG Pipeline

The application's core is a Retrieval-Augmented Generation pipeline:

1. Upload: A user uploads source files (e.g., a PDF textbook chapter, a lecture audio). The files are stored and associated with a specific lesson.
2. Ingestion: When ingestion is triggered, the backend processes each file:
   • It extracts raw text (from PDFs, DOCX), transcribes audio, or performs OCR on images.
   • The extracted text is split into smaller, manageable chunks.
   • These chunks are converted into vector embeddings (numerical representations) using a sentence-transformer model.
   • The embeddings are stored in a lesson-specific ChromaDB collection.
3. Generation (Retrieval + Augmentation):
   • When a user requests a lesson note, a prompt is created.
   • For targeted generation, the system searches the vector database for the most relevant chunks of information related to the user's prompt.
   • For full lesson generation, the system retrieves all chunks of information.
   • This retrieved content (the "context") is combined with the user's prompt and sent to a Large Language Model (like Google Gemini).
   • The LLM generates the lesson note based on the rich context provided, and the result is saved and displayed to the user.

Tech Stack

• Backend: Python, Flask, SQLAlchemy
• Database: MariaDB / MySQL
• Vector Database: ChromaDB (local persistent storage)
• RAG & AI: google-generativeai, sentence-transformers, langchain (for text splitting)
• Frontend: Vanilla JavaScript (ES6), Tailwind CSS, Marked.js (for Markdown rendering)
• Deployment: Docker & Docker Compose (optional, for containerized setup)

---

Installation and Setup

You can run this project either locally on your machine or using Docker.

Method 1: Local Machine Setup (Recommended)

Prerequisites:

• Python 3.9+
• A running MariaDB or MySQL server
• ffmpeg for audio processing (pkg install ffmpeg on Termux, brew install ffmpeg on macOS, sudo apt-get install ffmpeg on Debian/Ubuntu)

1. Clone & Setup Environment:

• Create a .env file from the example: cp .env.example .env
• Edit the .env file:
  • Add your GEMINI_API_KEY.
  • Set DATABASE_URL to point to your local database. Example:

    DATABASE_URL="mysql+mysqlconnector://YOUR_USER:YOUR_PASSWORD@127.0.0.1:3306/lesson_generator"
    

2. Setup Database:

• Log into your MySQL/MariaDB server as a root/admin user.
• Create the database and grant permissions:

  CREATE DATABASE lesson_generator;
  GRANT ALL PRIVILEGES ON lesson_generator.* TO 'YOUR_USER'@'localhost';
  FLUSH PRIVILEGES;

3. Install Dependencies & Run:

• Install Python packages:

  pip install -r backend/requirements.txt

• Run the Flask application:

  python backend/app.py

4. Access the App:

• Open your browser and go to http://localhost:8000.

Method 2: Docker Setup

Prerequisites:

• Docker
• Docker Compose (v2, i.e., docker compose)

1. Setup Environment:

• Create a .env file from the example: cp .env.example .env
• Edit the .env file with your GEMINI_API_KEY and any desired changes to the default MySQL credentials.

2. Build and Run Containers:

docker compose up --build -d

3. Access the App:

• Open your browser and go to http://localhost:8000.
• The database will be running in a container and accessible to the Flask app at the hostname db.

---

API Endpoints

All endpoints are prefixed with /api.

Method	Endpoint	Description
GET	/lessons	Get a list of all lessons.
POST	/lessons	Create a new lesson.
GET	/lessons/<id>	Get full details for a single lesson.
PUT	/lessons/<id>	Update a lesson's title or content.
DELETE	/lessons/<id>	Delete a lesson and all its associated data.
GET	/lessons/<id>/sources	Get a list of uploaded sources for a lesson.
POST	/lessons/<id>/sources	Upload a new source file.
DELETE	/lessons/<id>/sources/<source_id>	Delete a source file.
POST	/lessons/<id>/ingest	Trigger the ingestion process for uploaded sources.
POST	/lessons/<id>/generate	Generate/refine content with a simple prompt.
POST	/lessons/<id>/generate-full	Generate a full lesson from all ingested content.