grpo/docker-compose.yml

version: '3.8'

services:
  # Training service
  grpo-training:
    build:
      context: .
      target: training
    image: hanzoai/grpo:training
    container_name: grpo-training
    runtime: nvidia
    environment:
      - CUDA_VISIBLE_DEVICES=0
      - WANDB_API_KEY=${WANDB_API_KEY}
      - HANZO_API_KEY=${HANZO_API_KEY}
    volumes:
      - ./data:/app/data
      - ./models:/app/models
      - ./outputs:/app/outputs
      - ./config:/app/config
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: 1
              capabilities: [gpu]
    command: >
      python3 src/train.py
      --config /app/config/grpo_config.yaml
      --hanzo

  # Development environment
  grpo-dev:
    build:
      context: .
      target: development
    image: hanzoai/grpo:dev
    container_name: grpo-dev
    runtime: nvidia
    environment:
      - JUPYTER_ENABLE_LAB=yes
      - CUDA_VISIBLE_DEVICES=0
    volumes:
      - .:/app
      - ~/.cache:/root/.cache
    ports:
      - "8888:8888"  # Jupyter
      - "6006:6006"  # TensorBoard
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: 1
              capabilities: [gpu]
    command: >
      jupyter lab
      --ip=0.0.0.0
      --port=8888
      --no-browser
      --allow-root
      --NotebookApp.token=''

  # Inference API service
  grpo-inference:
    build:
      context: .
      target: inference
    image: hanzoai/grpo:inference
    container_name: grpo-inference
    runtime: nvidia
    environment:
      - MODEL_PATH=/app/models/production/latest
      - CUDA_VISIBLE_DEVICES=0
    volumes:
      - ./models:/app/models:ro
    ports:
      - "8000:8000"
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: 1
              capabilities: [gpu]
    healthcheck:
      test: ["CMD", "curl", "-f", "http://localhost:8000/health"]
      interval: 30s
      timeout: 10s
      retries: 3
      start_period: 40s

  # Daily training cron job
  grpo-daily-training:
    build:
      context: .
      target: training
    image: hanzoai/grpo:training
    container_name: grpo-daily-training
    runtime: nvidia
    environment:
      - CUDA_VISIBLE_DEVICES=0
      - HANZO_API_KEY=${HANZO_API_KEY}
    volumes:
      - ./data:/app/data
      - ./models:/app/models
      - ./config:/app/config
      - ./logs:/app/logs
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: 1
              capabilities: [gpu]
    command: >
      sh -c "while true; do
        echo 'Running daily training at' $$(date);
        python3 scripts/daily_training.py --config config/daily_training.yaml;
        echo 'Sleeping for 24 hours...';
        sleep 86400;
      done"

  # Redis for caching
  redis:
    image: redis:7-alpine
    container_name: grpo-redis
    ports:
      - "6379:6379"
    volumes:
      - redis-data:/data
    command: redis-server --appendonly yes

  # PostgreSQL for metadata
  postgres:
    image: postgres:15-alpine
    container_name: grpo-postgres
    environment:
      - POSTGRES_DB=grpo
      - POSTGRES_USER=grpo
      - POSTGRES_PASSWORD=${POSTGRES_PASSWORD:-grpo123}
    ports:
      - "5432:5432"
    volumes:
      - postgres-data:/var/lib/postgresql/data

  # MinIO for model storage
  minio:
    image: minio/minio:latest
    container_name: grpo-minio
    environment:
      - MINIO_ROOT_USER=${MINIO_ROOT_USER:-minioadmin}
      - MINIO_ROOT_PASSWORD=${MINIO_ROOT_PASSWORD:-minioadmin}
    ports:
      - "9000:9000"
      - "9001:9001"
    volumes:
      - minio-data:/data
    command: server /data --console-address ":9001"

volumes:
  redis-data:
  postgres-data:
  minio-data:

networks:
  default:
    name: grpo-network