I build the system
to enable the scale.

My company, HERE, runs on four surfaces — a member app on iOS and Android, a website, a CRM, and a portal for restaurants — over a single API. Run by hand, that’s four places to watch and a founder in the loop on every booking, payment, reminder and no-show. A business like that can’t scale, and it can never run without the person sitting at the centre of it.

So I built it to run itself. One event-driven backbone where a booking, a payment or a cancellation fires the next step automatically — reminders sent, waitlists advanced, capacity recalculated, feedback requested, the team alerted — with reconcilers that quietly repair the state when a step fails. No dashboards to check, no manual tasks, no me in the loop. The whole operation runs across 74 cities and 10 currencies while I sleep.

4 surfaces, one backbone · 74 cities · 0 manual steps in the daily loop

HERE members subscribe from two places — the app and the web app — and both have to resolve to the same billing truth. Subscriptions, annual-to-monthly switches, coupons, dunning, cancellations and reinstatements: every one of them a moment for the two clients to disagree, for a webhook to arrive twice, or for a member to be charged the wrong amount in the wrong currency.

I built a single Stripe billing spine that both surfaces call through — one source of truth for every subscription, schedule and coupon, across ten currencies. Every mutation is idempotency-keyed, so a retry or a duplicated webhook can never double-charge or double-grant; every outcome is reconciled back against Stripe, so the database and the ledger can’t drift apart; and each flow is fenced and self-healing, detaching cleanly when a step fails mid-write. Whether a member checks out on their phone or in a browser, they get the identical, correct result — and I prove it against a Stripe test clock before a single change reaches production.

app + web · one ledger · every mutation idempotency-keyed · 0 errors by design

Both of the consumer businesses I’ve built rest on the same idea: bringing together people who genuinely connect. Underneath sits a compatibility methodology developed over years of work with philosophers and psychologists — rigorous, principled, and expressed in ideas rather than numbers. To act on it at scale, that methodology had to become computable: a scoring system precise enough to rank compatibility, without losing any of the depth years of research had built into it.

I translated the methodology into a mathematically defined scoring system — then validated it through extensive testing. Across successive beta rounds I ran competing scoring models against real respondents, measuring not just whether the matches held up, but whether the questions read clearly and the results felt true to the people answering them. Both halves were tuned in the same loop: the questions and answers on the surface, and the weightings underneath. What came out stays faithful to the research and stands up as mathematics — a considered model of human connection, made measurable, repeatable and accurate.

one methodology → one score · tuned over multiple beta rounds · mathematically curated, never random.