StudyPal Assistant

A modular Retrieval Augmented Generation (RAG) assistant for Class 12 STEM subjects, combining semantic search over textbooks/chapters and conversational AI. Designed for PDF-based knowledge ingestion, vector storage (ChromaDB), and interactive Q&A with Streamlit.

Features

• Conversational AI: Uses Groq LLM and streaming chat memory for contextual, multi-turn Q&A.
• Document Ingestion: PDF chapters vectorized using HuggingFace embeddings.
• Semantic Search: ChromaDB allows deep, similarity-based text retrieval.
• YouTube Integration: Suggests relevant videos for study topics.
• Extensible UI: Streamlit front-end with dropdowns for subjects/chapters.

Project Structure

StudyPal-assistant/
├── data/
│   └── class_12/
│       └── biology/
│           ├── 1.Life_Process.pdf
│           └── 2.Reproduction.pdf
├── vector_db/
├── chapters_vector_db/
├── src/
│   ├── main.py                # Streamlit web app
│   ├── chatbot_utility.py     # Chapter list logic
│   ├── get_yt_video.py        # YouTube search integration
│   ├── vectorize_book.py      # Vectorization pipeline
│   └── vectorize_script.py    # CLI for vectorization
├── .env
└── README.md

Quick Start

1. Clone and Install

git clone https://github.com/MLayush-dubey/StudyPal-assistant.git
cd StudyPal-assistant
python -m venv .venv
.venv\Scripts\activate      # On Windows
pip install -r requirements.txt

2. Set Environment Variables

Create a .env file in the root:

GROQ_API_KEY=YOUR_API_KEY
CLASS_SUBJECT_NAME=biology
DEVICE=cuda      # or 'cpu' if you have a low-end/CPU-only setup

3. Add PDFs

Place your subject PDFs inside:

data/class_12/biology/

4. Vectorize

cd src
python vectorize_script.py

5. Launch the App

streamlit run main.py

Note

I have only included one subject (Biology) with two chapters in my database, since I use a low-end GPU.
If you have a more capable GPU, add more subjects (e.g. Physics, Chemistry) and chapters for broader experimentation and practice.
In the Streamlit app, you’ll see Chemistry and Physics in the dropdown—these are currently empty placeholders. Feel free to add PDFs for these subjects if you want to expand or test the system.

Main Components

• main.py: Streamlit chat UI, retrieval chain setup, search & memory features, YouTube recommendations.
• chatbot_utility.py: Chapter list logic per subject.
• get_yt_video.py: Fetches top-3 related YouTube videos using keywords.
• vectorize_book.py / vectorize_script.py: Prepares the vector DB for semantic search.

Hardware Notes

• Supports both CPU and CUDA-based inference.
• For large-scale ingestion/vectorization, a more powerful GPU is recommended.

Feel free to copy, edit, or extend this according to your evolving project!