compose.yaml

version: '3.8'

services:
  call-center-ai:
    build: .
    container_name: call_center_engine
    restart: always
    ports:
      - "8000:8000"
    environment:
      # Team will paste the token here or in a .env file
      - HF_TOKEN=${HF_TOKEN}
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: 1
              capabilities: [gpu]
    volumes:
      # Optional: Persist models so they don't re-download on restart
      - huggingface_cache:/root/.cache/huggingface

volumes:
  huggingface_cache: