Deployment Guide: Docker Compose Stack

# 1. Clone & config
git clone https://github.com/vtnguyen04/phoenix_ML.git
cd phoenix_ML
cp .env.example .env

# 2. Start infrastructure services
docker compose up -d

# 3. Start Airflow (optional)
docker compose -f docker-compose.airflow.yaml up -d

# 4. Generate data & train models
uv run python scripts/generate_datasets.py
uv run python examples/credit_risk/train.py

# 5. Verify
curl http://localhost:8001/health
# → {"status": "healthy", "version": "1.0.0"}

# ── Database ──
DATABASE_URL=postgresql+asyncpg://phoenix:phoenix@postgres:5432/phoenix
TEST_DATABASE_URL=sqlite+aiosqlite:///./phoenix_test.db

# ── Redis ──
REDIS_URL=redis://redis:6379

# ── Kafka ──
KAFKA_URL=kafka:9092

# ── MLflow ──
MLFLOW_TRACKING_URI=http://mlflow:5000

# ── Model Config ──
DEFAULT_MODEL_ID=credit-risk
DEFAULT_MODEL_VERSION=v1
MODEL_CONFIG_DIR=model_configs
INFERENCE_ENGINE=onnx

# ── Monitoring ──
MONITORING_INTERVAL_SECONDS=30
DRIFT_THRESHOLD=0.3

# ── Storage ──
ARTIFACT_STORAGE_DIR=./artifacts
CACHE_DIR=./cache

# ── Ports ──
API_PORT=8001
FRONTEND_PORT=5174

make up          # docker compose up -d
make down        # docker compose down
make build       # docker compose build
make restart     # down + up
make logs        # docker compose logs -f
make ps          # docker compose ps
make clean       # down -v (remove volumes)
make test        # uv run pytest
make api-logs    # docker compose logs -f phoenix-api
make frontend-logs  # docker compose logs -f phoenix-frontend

# Generate data + train all models
uv run python scripts/generate_datasets.py
uv run python examples/credit_risk/train.py
uv run python examples/house_price/train.py
uv run python examples/fraud_detection/train.py
uv run python examples/image_classification/train.py
uv run python examples/sentiment/train.py

# Run end-to-end simulation
uv run python scripts/simulate_pipeline.py --fast

# Or trigger via Airflow UI (http://localhost:8080)

# Install via Helm
helm install phoenix-ml deploy/helm/phoenix-ml/

# Custom values
helm install phoenix-ml deploy/helm/phoenix-ml/ \
  --set replicaCount=3 \
  --set image.tag=v1.0.0 \
  --set resources.limits.memory=2Gi

# Upgrade
helm upgrade phoenix-ml deploy/helm/phoenix-ml/

# Uninstall
helm uninstall phoenix-ml

deploy/helm/phoenix-ml/
├── Chart.yaml              # name: phoenix-ml, version: 0.1.0
├── values.yaml             # Default: 1 replica, 512Mi memory
└── templates/
    ├── deployment.yaml     # Pod spec, health probes, env vars
    ├── service.yaml        # ClusterIP on port 8000
    ├── ingress.yaml        # External access, TLS
    └── hpa.yaml            # Auto-scale: 1-10 replicas, 80% CPU target

# Apply all migrations
uv run alembic upgrade head

# Create new migration
uv run alembic revision --autogenerate -m "add new table"

# Rollback last migration
uv run alembic downgrade -1

# View current version
uv run alembic current

models
├── id (VARCHAR, PK)
├── version (VARCHAR, PK)
├── uri (VARCHAR)
├── framework (VARCHAR)
├── stage (VARCHAR: champion/challenger/retired/archived)
├── metadata_json (JSONB)
├── metrics_json (JSONB)
├── is_active (BOOLEAN)
└── created_at (TIMESTAMP)

prediction_logs
├── id (UUID, PK)
├── model_id (VARCHAR)
├── model_version (VARCHAR)
├── features_json (JSONB)
├── result (FLOAT)
├── confidence (FLOAT)
├── latency_ms (FLOAT)
├── ground_truth (FLOAT, nullable)
└── timestamp (TIMESTAMP)

drift_reports
├── id (UUID, PK)
├── model_id (VARCHAR)
├── feature_name (VARCHAR)
├── method (VARCHAR: ks/psi/chi2)
├── score (FLOAT)
├── is_drifted (BOOLEAN)
├── threshold (FLOAT)
└── timestamp (TIMESTAMP)