Founding Revenue Systems Engineer

Why This Role Exists
You will own two revenue engines: the one that pays for everything today, and the one we are switching on now.

The first engine already works. Hundreds of thousands of stores, millions of shoppers, real money moving every day — it funds everything else we build. A few engineers and the founder run it, and it holds far more upside than its owners have hours.

The second engine barely exists. Our developer API goes paid: AI agents and outside developers buying verified-commerce data by the key. Paid keys need a billing spine — metering, quota, spend caps, tiered plans, invoices a customer can trust. Today that spine has no owner.

This seat owns both. You steward the money that already flows, and you build the money that is about to. You decide what to build, and you are measured in dollars.

The System You'll Need to Model
Usage-based billing as a correctness problem. Metering, rating, quota, and invoicing for a paid API: every call counted once, capped correctly, billed exactly. The craft is idempotent event ingestion, exactly-once counting under retries, and spend-cap enforcement that fails closed. The meter is the product — an error in the meter is an error on the invoice.
Financial reconciliation across sources you do not control. Dozens of affiliate networks report what we earned, each on its own schema, its own lag, its own reversal and clawback rules. A bug in this path does not crash anything — it quietly leaks five figures before anyone notices. Ledger correctness is the invariant the whole company stands on.
Attribution across long async windows. Every dollar traces back to one fragile click ID that has to survive an async gap of hours to months, across dozens of networks that each name and format it differently. This is attribution infrastructure with payments-grade stakes: get it wrong and you do not slow a page — you misstate the revenue.
Billing for machine customers. The next buyers of commerce data are AI agents, not people — the way Google once chose which pages to index, agents now choose which APIs to call. Keys are the new accounts. The spine has to treat a robot as a first-class paying customer: authenticated, metered, capped, and invoiced.
Cortex, the brain the company runs on. Every operator works through governed AI sessions inside Cortex, our shared AI brain — the same system we sell as a product family — and the substrate answers its own questions from more than 8,600 internal documents. Agents write much of the code; humans own design, failure modes, and verdicts. The company evolves at that pace, and you model where it is going; nobody hands you a brief.
If reading that energizes you, keep going. If it feels overwhelming or underspecified, this isn't the right fit.

What You Will Own
The billing and metering spine, end to end. Keyed developer access, usage metering, quota, spend caps, tiered plans, and the invoice that proves it all right — the machinery that turns a free API into paid revenue. You architect it and ship it on our stack (TypeScript and Node services on PostgreSQL), and when the commercial seat signs a developer to a paid tier, your spine is what delivers the promise. On a resume this is usage-based billing infrastructure, built from zero and run in production.
The commission engine. The multi-network ingestion, attribution, and reconciliation that carry the existing revenue line of around $22M a year. It runs today; you take it further than anyone has had the hours to — tighter reconciliation, faster leak detection, cleaner ledger invariants. This is production financial infrastructure with real dollars on every code path.
Two public cutovers. Anonymous API access ends August 15. Legacy free codes expire September 30. The keying and metering have to be live before the free lane closes, and both cutovers are yours to land.
The seat itself. Agents write most of the code here; the scarce thing is judgment — reading a system well enough to say why the code is right — and that verdict is yours, measured in dollars moved, not features shipped. Inside your first quarter you co-sign a seat charter: one machine-checkable number that proves the seat is working, plus a written split of what you decide alone and what you bring to the founder first. The craft you must own walking in is payments-grade correctness: idempotency, reconciliation, ledger integrity. What you grow into here: billing where the customers are machines, and coding agents run as a production workforce.

Who You Are
You reason in invariants, failure modes, and tradeoffs, and you tie each one to the dollar it moves. You can read a system you have never seen well enough to sketch where it leaks money the same day. When a number looks wrong, you instrument the pipeline and find the mechanism — reconciliation drift is something you chase to a root cause, never noise you learn to tolerate.

Agents are your production system. You direct them and you verify what comes back — you can do this job by hand and prove it, and that mastery is exactly what lets you trust or reject what an agent hands you. You treat agent output as something you check, never something you accept on faith. The expensive thing here is a redo cycle, never the compute.

You have owned a system where the outcome was real money — payments, billing, metering, an affiliate or commission ledger, attribution, fraud scoring, or a marketplace transaction path at scale. You can point at it and explain the mechanism that moved the number, not just show the chart. You know why idempotency keys exist because you have paid for their absence. Where you did it and what you studied matter far less than that you built the money path and watched it hold under load. That is the transfer we want.