Knowledge Assistant API

A Django-based backend system that powers a Knowledge Assistant API using OpenAI LLM and FAISS for retrieval-augmented generation (RAG) from PDF, Markdown, or Text knowledge bases.

Features

• Upload and ingest knowledge base documents (PDF, Markdown, Text)
• Parse, chunk, and embed content using OpenAI
• Store embeddings in FAISS for semantic search
• Ask questions via API, with answers grounded in uploaded knowledge
• Admin interface for document management

Tech Stack

• Python 3.x
• Django + Django REST Framework
• OpenAI API (for embeddings and LLM)
• FAISS (vector search)
• PyPDF2, python-docx, markdown (parsing)

Setup

1. Clone and install dependencies:

   python3 -m venv venv
   source venv/bin/activate
   pip install -r requirements.txt

2. Set OpenAI API Key:

   export OPENAI_API_KEY=your-openai-key

3. Run migrations:

   python manage.py makemigrations
   python manage.py migrate

4. Create superuser (for admin):

   python manage.py createsuperuser

5. Run the server:

   python manage.py runserver

API Endpoints

1. Upload Document

• POST /api/upload-document/
• Form Data: file (PDF, .md, .txt)
• Response: Document metadata

2. Ask Question

• POST /api/ask-question/
• JSON: { "question": "What is the use of mitochondria?" }
• Response:

  {
    "answer": "The mitochondria is known as the powerhouse of the cell...",
    "sources": ["Document.pdf - Chunk 3"]
  }

Admin

• Visit /admin/ to manage documents and chunks.

Notes

• Only context from uploaded documents is used to answer questions (RAG).
• Caching and advanced prompt engineering can be added for optimization.

Example Knowledge Base

• Upload a Science Class IX PDF or Markdown file to test. # Final-LLm

Final-LLm

Final-LLm