Matteen Labs
OPEN POSITION: Lead AI Product Engineer
Full-time role, entered through a paid trial build. Starting immediately. US preferred, open to exceptional candidates anywhere.
The problem
It works. It doesn't scale past a few dozen people.
We run a decision intelligence practice. Members get a private AI that holds their complete chart data across four systems — and answers questions about how they specifically make decisions.
Every instance is built by hand. Our founder creates each member's system personally. The output quality is the entire product.
We need the engine that generates a member's instance from their birth data, at the same quality, on our own infrastructure.
More about the practice at matteen.com.
The platform
One engine, three surfaces
This isn't one app. It's a platform with three consumers, and the architecture has to know that on day one.
Building single-product and retrofitting the second tier later is the exact failure we're hiring to avoid.
The work
The parts that are actually hard
If you have opinions about any of these, we want to hear them on the first call.
The process
How this works
If that's more than you want to do to apply, we understand, and this probably isn't the right fit.
Infrastructure
The stack
Auth runs on FusionAuth. Subscriptions and CRM run on GoHighLevel, alongside WordPress and a few other systems. Entitlement checks hit GoHighLevel directly — that's what keeps us independent of third-party platform control.
We currently run on custom GPTs. We'd prefer to build on Anthropic's Claude, which means part of the early work is proving quality holds through that move. If you think that's the wrong call, make the case.
Chart computation already exists and you'd inherit it. The math is solved. The architecture isn't.
Timeline
Ships to both stores, this year
That's four months. It means v1 ships narrow — premium tier only, architected for the rest. App Store review is part of the timeline, rejections included.
If you think that date is wrong, we'd rather hear it in your first week than in November.
The offer
How it starts
This is a full-time role. That's what we're hiring for and that's where it goes.
You get there through a paid trial build — usually two weeks, sometimes a smaller project first. Real work, real money, a defined deliverable. Not an interview exercise, and not a contract that might convert someday.
We do it this way because nobody can tell from a résumé whether they'd want to work with someone for the next three years, and neither can you. Two weeks tells both of us. If it isn't a fit, the work stands on its own and nobody owes anybody a longer conversation.
The shape of the trial depends on the person. Some people we'll want to see on something small first. Others we'll take straight into the full build. We'll tell you which on the call.
We want someone starting now, not in six weeks.
The fit
Who this is for
- You've shipped LLM-backed product to real paying users. Not demos, not internal tools.
- You've shipped a mobile app to the App Store or Play Store — or you can make a credible case for why you'll handle it.
- You have opinions about context engineering and can explain why an approach fails, not just that it does.
- You supply architecture. You don't wait to be handed it.
- You're comfortable telling us on day three that our scope is wrong, if it is.
You don't need to know anything about Human Design or astrology, and nobody expects you to learn it. The domain is a data model. The hard part is architecture and quality at scale.
You'd be the most senior technical person on this product and you'd own it end to end. There's an AI developer already shipping and a programmer on infrastructure. You work alongside them, not above them. Nobody reports to you.
Step one
Apply
Four things: your HumanCharts profile, your resume, links to work you've shipped, and a two-minute video on why this project interests you.
Every field marked required is required. Video links only — unlisted YouTube, Loom, or Drive.
Our team reads every application. Nobody goes quiet — if it's a no, we'll tell you.