Docker¶

Full stack (recommended)¶

Run Postgres, Redis, the API server, and the arq worker — all in containers:

cp .env.example .env
bash dev.sh

dev.sh builds the inference-engine image and starts all services via Docker Compose. See Docker Compose for a full breakdown of services, volumes, and profiles.

Dockerfile¶

The project ships a two-stage Dockerfile:

builder — installs Python dependencies into .venv via uv sync --frozen --no-dev
runtime — copies .venv and app/ into a clean python:3.12-slim image under a non-root appuser

Build and run standalone (falls back to SQLite + in-process async — no Postgres/Redis needed):

docker build -t inference-engine .
docker run -p 8000:8000 \
  -e API_KEYS="key1:tenant_a:predict,read_models" \
  -e ENV=production \
  inference-engine

Services and ports¶

Service	Host port
API server	`8000`
Postgres	`15432`
Redis	`6379`

Observability stack¶

bash dev.sh --observability

Starts Prometheus (9090) and Grafana (3000) alongside the core services. To add observability to an already-running stack:

docker compose --profile observability up -d

See Docker Compose for the full reference.