Deevnet IoT as a Service #

The Mobile Factory is a portable IoT as a Service platform, bundled in the base of a modular toolkit. Set it up on site, and the people in the room get a predictable network for their devices and back-end services such as MQTT messaging, logs and metrics dashboards, all declared from their own Terraform, with no public cloud needed because this platform is its own mini-cloud.

Why I built this #

I’m a member of the Columbus Arduino and Raspberry Pi Enthusiasts. I built the Mobile Factory for my own projects: I wanted to build things on site at meetups, and that meant bringing my own network. Then eventually, as the project evolved, new ideas came along and I wanted to make it something everyone else could potentially use too.

Like a typical engineer, I couldn’t just build the thing. First I had to build the thing that builds the thing. So the network, the services and the images it runs are all automated and built from code, and everything can be rebuilt from scratch. Somewhere along the way this project became a reference implementation for infrastructure automation.

It’s all bundled in the base of a Bauer modular toolkit, so the device components, a portable electronics workbench and supplies come on site with it, ready for hardware hacks and prototyping.

Come build with it #

If you’re local to Columbus, Ohio, come to a CARPE meetup and get hands-on. Bring your own breadboard, microcontrollers and sensors, or borrow devices at the meetup; the only hard requirement is your laptop ( Before You Start says what to have on it). Either way, you can rapidly prototype a multi-device project with the backend services already running:

• Messaging and logging are already decided, with an opinionated approach: MQTT with an account per device, and logs you can chart in your own dashboards. You spend the evening on your devices and what they do, not on standing up a broker.
• No public cloud. Your devices talk to services in the room. There’s no cloud account to sign up for, and nothing leaves the site.
• Bring a microSD card, and you can take your backend home with you, to run on a Pi of your own ( Convert a Tenant to a Pi Image).

If you’re not local, the code and these docs are open. Use them and adapt them for your own lab, or, these days, have your favorite AI agents adapt them for you.

Features #

A tenant declares what it needs in its own Terraform and gets:

• A network for its devices. Its own Wi-Fi key on the IoT network, issued to the tenant by the API, with no one touching the controller.
• MQTT. Broker accounts scoped to the tenant’s own topics, and a registry of its devices.
• Logs and dashboards. Its workloads’ and devices’ logs in a store only it can read, and its own Grafana organization to chart them.
• Isolated workloads. Its own network, DNS zone and state storage, rebuilt from its code.

What makes it a factory #

• It builds everything from code: its network, its services and the images it runs, and it works offline once it is set up.
• It produces what tenants take away: the take-home Pi kit, which reproduces the same services on a single Raspberry Pi, and the prebuilt provider and tools a developer needs on their own laptop.
• It travels. Its address space and DNS zone move with the toolkit, so it is the same factory wherever it is set up, with no renumbering.

In the toolkit’s base: router, switch, wireless AP, two Proxmox hypervisors and Raspberry Pis. The rest of the toolkit carries the breadboards, components and tools for prototyping devices.

Explore the Documentation #

A factory of sites and tenants #

This site documents the Mobile Factory, the mobile instance of Deevnet. The Mobile Factory builds sites and tenants. Its sites are itself and, next, the home site; its tenants are the devices and services it creates, deployed at whichever site hosts them. Architecture, Standards and Policies are written for any Deevnet site. Implementation & Tooling, the Runbook, the records and the Roadmap are the factory’s, including what it deploys at other sites.

The patterns, automation and documentation are meant to be adaptable: the standards and architecture can be applied to your own collection of devices and networks.

Documentation Philosophy #

This documentation exists to:

• seed context to AI agent tooling,
• make intent explicit,
• prevent knowledge from living only in someone’s head,
• and ensure future changes remain coherent.

If something “works” but violates these documents, it is considered incorrect.