services:
  ollama:
    image: ollama/ollama
    volumes:
      - ollama:/root/.ollama
    restart: unless-stopped
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: all
              capabilities: [gpu]
    healthcheck:
      test: ollama --version && ollama ps || exit 1
      interval: 60s
      retries: 5
      start_period: 20s
      timeout: 10s
    networks:
      - nointernet

  ollama_provision:
    build:
      dockerfile: src/ollama_provision/Dockerfile
    environment:
      - MODELS=nomic-embed-text,llama3.1:8b,qwen2.5:7b,gemma3:12b-it-qat,gemma3:27b-it-qat,gpt-oss:20b
    volumes:
      - ollama:/root/.ollama
    restart: no
    depends_on:
      ollama:
        condition: service_healthy
        restart: true
    links:
      - ollama
    networks:
      - internet

  speaches-ai:
    image: ghcr.io/speaches-ai/speaches:latest-cuda
    restart: unless-stopped
    volumes:
      - hf-hub-cache:/home/ubuntu/.cache/huggingface/hub
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: all
              capabilities: [gpu]
    healthcheck:
      test: curl --fail http://localhost:8000 || exit 1
      interval: 60s
      retries: 5
      start_period: 10s
      timeout: 10s
    networks:
      - nointernet

  speaches-ai_provision:
    build:
      dockerfile: src/speaches-ai_provision/Dockerfile
    volumes:
      - hf-hub-cache:/home/ubuntu/.cache/huggingface/hub
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: all
              capabilities: [gpu]
    restart: no
    networks:
      - internet

  aichat:
    build:
      dockerfile: src/aichat/Dockerfile
    restart: unless-stopped
    environment:
      - API_KEY=${LLM_API_KEY}
    depends_on:
      - ollama
    links:
      - ollama
    healthcheck:
      test: timeout 10s bash -c ':> /dev/tcp/127.0.0.1/8000' || exit 1
      interval: 30s
      timeout: 15s
      retries: 3
    networks:
      - nointernet

  nginx:
    image: nginx
    volumes:
      - ./src/nginx/nginx.conf:/etc/nginx/templates/nginx.conf.template
      - ./src/nginx/htpasswd:/etc/nginx/.htpasswd
    environment:
      - NGINX_ENVSUBST_OUTPUT_DIR=/etc/nginx
      - API_KEY=${LLM_API_KEY}
    depends_on:
      - ollama
      - speaches-ai
      - aichat
    links:
      - ollama
      - speaches-ai
      - aichat
    ports:
      - "127.0.0.1:10434:10434"
      - "127.0.0.1:8000:8000"
      - "127.0.0.1:8001:8001"
    healthcheck:
      test: curl --fail http://localhost:11434 || exit 1
      interval: 10s
      retries: 5
      start_period: 10s
      timeout: 10s
    restart: unless-stopped
    networks:
      - internet
      - nointernet

volumes:
  ollama:
  hf-hub-cache:

networks:
  internet:
    internal: false
  nointernet:
    internal: true