🧠 Intelligence System for Academic Notes

A Context-Bound QA System that answers student questions only based on uploaded academic notes. It uses RAG (Retrieval Augmented Generation) to ensure grounded answers and strict hallucination control.

🚀 Features

• Text Chunking: Splits large academic PDFs into manageable logical chunks.
• Semantic Search: Uses Local Embeddings (HuggingFace) to find relevant content without API rate limits.
• Context-Bound AI: Answering engine (Llama-3 via Groq) is strictly prompted to admit ignorance if data is missing.
• Hallucination Control: Returns "Information not available in the notes" for out-of-scope questions.
• Source Attribution: Cites the specific page number and file name for every answer.

🛠️ Tech Stack

• Backend: Flask (Python)
• Vector DB: ChromaDB (Persistent storage)
• LLM: Llama-3.3-70b (via Groq)
• Embeddings: all-MiniLM-L6-v2 (HuggingFace)
• Framework: LangChain

🏃‍♂️ How to Run

1. Clone the Repository

   git clone <your-repo-url>
   cd academic-notes-ai

2. Install Dependencies

   pip install -r requirements.txt

3. Setup Environment Variables Create a .env file in the root directory:

   GROQ_API_KEY=gsk_your_key_here...

4. Run the Server

   python app.py

🔌 API Endpoints

1. Upload Notes (POST /upload)

Uploads a PDF or Text file to the knowledge base.

• Body: form-data -> key: files (File)

2. Ask Question (POST /ask)

Asks a question based on the uploaded notes.

• Body: raw JSON

  {
    "question": "What is the definition of mitosis?"
  }