azure-role-based-assistant

Backend implementation for a simple AI assistant using Azure OpenAI, Azure AI Search, and FastAPI.

This project demonstrates how to create an intelligent assistant, that responds to questions based on their role (e.g. HR, IT). It utilizes retrieval-augmented generation (RAG) technique to provide accurate answers from internal data.

Setup instructions

1. Clone repository:

git clone https://github.com/eetuhoo/azure-role-based-assistant.git
cd azure-role-based-assistant

2. Create virtual environment and activate it:

python -m venv venv
venv\Scripts\activate

3. Install dependencies:

pip install -r requirements.txt

4. Create .env file for environment variables:

AZURE_OPENAI_ENDPOINT=...
AZURE_OPENAI_API_KEY=...
AZURE_OPENAI_DEPLOYMENT=...
AZURE_OPENAI_API_VERSION=...
AZURE_SEARCH_ENDPOINT=...
AZURE_SEARCH_API_KEY=...
AZURE_SEARCH_INDEX=...

How to run

uvicorn main:app --reload

Test API endpoint /rag-ask in e.g. Swagger-documentation:

http://localhost:8000/docs
POST /rag-ask
{
  "role": "IT",
  "question": "Kuinka uusin salasanani?"
}

The backend should respond with a role-specific answer.