Deploy Your Code, Your Way — Introducing Arango BYOC

What if your custom ML model, API, or data-processing service could live right next to your data — natively authenticated, automatically routed, and fully isolated — without you managing any of that yourself? That’s exactly what Arango’s Bring Your Own Code (BYOCode) and Bring Your Own Container (BYOContainer) frameworks make possible.

The Container Manager — a specialized orchestration service within the Arango Contextual Data Platform — lets you deploy and run your own services right alongside your data, all orchestrated through the Arango Control Plane (ACP). Authentication, network isolation, telemetry, and license management are handled automatically, so your team can stay focused on building logic, not infrastructure.

Two Paths to Deployment

The platform gives you flexibility to meet your team where it is. Whether you want to hand off container-building entirely or keep full control of your runtime environment, there’s a path for you:

Getting Started with ServiceMaker

For the code-based path, Arango recommends ServiceMaker — a CLI tool that automates the entire preparation workflow. It handles dependency installation via uv, builds local images for testing, and packages everything into a deployment-ready archive. Built in Rust, it’s fast and opinionated in all the right ways.

git clone https://github.com/arangodb/servicemaker.git
cd servicemaker
cargo build --release
./servicemaker --input /path/to/your/project --name my-project-name

Key Requirements to Keep in Mind

Before you deploy, two networking rules are non-negotiable: your service must expose an HTTP server on port 8000, and it must handle requests at the root path (/). Here’s a minimal FastAPI setup that meets both:

from fastapi import FastAPI
import uvicorn

app = FastAPI()

@app.get("/")
def read_root():
    return {"status": "Service is running on Arango Container Manager"}

if __name__ == "__main__":
    uvicorn.run(app, host="0.0.0.0", port=8000)

Choose Your Deployment Scope

One of the most powerful BYOC features is choosing exactly where a service lives in your deployment topology:

Deploying via the UI — Five Steps

1. Navigate to Control Panel → Container Manager after logging in.
2. Select the Packages tab for code uploads, or Containers for Docker image URLs.
3. Drag and drop your .tar.gz, or paste your Docker image URL directly.
4. Provide a filename, semantic version (e.g. 1.0.0), service URL path, and base image. Multiple versions can run simultaneously.
5. Click Deploy Service. The Control Plane handles everything else.

Common Gotchas (and How to Fix Them)

ARM ARCHITECTURE MISMATCH

Running ServiceMaker on an M-series Mac? If it tries to pull an amd64 image, the build will fail. Fix: manually build the base image locally first:

cd servicemaker/baseimages
docker build -f Dockerfile.py13base -t arangodb/py13base:latest .

Then rerun ServiceMaker — it will use your local image and skip the pull.

OPTIONAL DEPENDENCIES WON’T PACKAGE

ServiceMaker doesn’t currently support uv sync –extra optional dependency groups. If your pyproject.toml uses extras, those packages won’t make it into the container. Fix: move all required dependencies into the main dependencies array rather than tucking them behind optional groups.