Environment Setup & Build

1. Environment Configuration

Create a .env file in the project root (copy from .env.example):

cp .env.example .env

The .env file contains API key, URLs, and processing configuration:

# API Authentication
API_KEY=your_api_key_here

# URLs Configuration
STREAM_URL=https://95.217.75.14:8443/stream
FLAG_URL=https://95.217.75.14:8443/api/flag
LOCAL_PREDICT_URL=http://127.0.0.1:8000/predict?store=1
FRONTEND_POST_URL=https://hackabaka.vercel.app/api/stream

# Processing Configuration
MAX_WORKERS=4
THRESHOLD=0.35

For local development, you can override URLs as needed (e.g., FRONTEND_POST_URL=http://localhost:3000/api/stream).

Note: The .env file is git-ignored for security. Never commit API keys or sensitive data.

2. Python Dependencies

Install all dependencies:

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

3. Train the AI Model

To train the fraud detection model on a CSV dataset:

python3 backend/train_model.py --input hackathon_train.csv --db history.db --output-model model.pkl --features backend/features.json

4. Frontend Dependencies

Build the Next.js frontend:

cd frontend
npm install
npm run build

Launch Backend & Frontend

The following terminals and their respective commands should be run in the given order.

Terminal 1 - Frontend Dashboard:

cd frontend
npm run start

Terminal 2 - Backend API:

cd backend
python3 -m uvicorn fraud_api:app --host 127.0.0.1 --port 8000

Terminal 3 - Stream Processor:

python3 sse_to_predict.py

The dashboard will be available at http://localhost:3000