I build and own products end-to-end — from the problem to what ships.

The first six months weren’t code. Fidamy sells device insurance through retail partners rather than directly to consumers, so the real question was what to build — which flows customers actually understood, what our investor (NN) would approve, and what our insurance licensing allowed. In a regulated market, that discovery is the work, not a delay before it.

The purchase flow taught me something about architecture and timing. The step count kept moving — six at MVP, then five, then four — as we cleared each version with NN and the regulator. Next.js routing is file-based, so pinning each step to a URL while they churned would have locked me in too early; I drove the wizard from client-side state to stay flexible. That had a cost — a race condition loading step data caused hydration errors. Once the design stabilized at four approved steps, I re-architected to file-based per-step routing, each step loading deterministically. It killed the race condition, made the flow resumable and shareable, and took checkout conversion from ~20% to ~67%.

The other hard problem was speed — and cost. Our pricing engine (Peak3) had no bulk-quote endpoint and billed per API call, so four quotations meant four slow, metered requests. Its pricing is deterministic, so I cached quotes in our own price map and built the bulk endpoint it never offered: quote loads dropped from ~10s to under 100ms, and repeat pricing stopped costing us anything.

Some decisions were about the business, not the code. Moving payments from Stripe to Twikey wasn’t a technical flex — iDEAL and SEPA direct debit are how the Netherlands pays, and direct-debit mandates were a hard requirement for an insurer. What started as a purchase form had become a distribution platform.

I was brought in on a referral, not a job posting. I’d just led the enterprise integration that migrated Freedom Mobile’s customer base into Rogers’ systems after the acquisition, and the person who watched me deliver it vouched for me directly to Droobi’s CTO — ex-McKinsey, ex-Microsoft. That’s a pattern I’ve come to rely on: the hardest roles reach me because someone has already seen me handle complexity firsthand.

My title was “Technical Lead,” but I was running the engineering function. A ten-person Venture Dive team reported to me across four disciplines; I set the standards and the delivery cadence, had a direct hand in what we paid the vendor and how we allocated engineers against budget, and still built the product’s web frontend myself. As the only person on-site, I was the single point where product, delivery, and cost all met.

The stakes were clinical and real. We served the diabetes patients of two of Doha’s major hospitals — Hamad Medical Corporation and Al Ahli — with iOS and Android apps for patients, dashboards for the clinicians tracking them, and integrations with glucose monitors, scales, and wearables feeding live metrics back to care teams.

The hardest constraint wasn’t clinical — it was where the data was allowed to live. Qatari law bars government-hospital health data from the public cloud, so AWS and Azure were off the table. I made the call to host on a custom Ooredoo data center, brought in a specialist firm to build it, and steered the architecture around on-shore hosting. That one requirement shaped the entire platform — and we still shipped, including exposure-tracking and in-app video consultations on a compressed timeline when COVID hit.

Owning a yacht is out of reach for almost everyone, but the thing people actually want isn’t the asset — it’s the time on the water. So we let people buy a fractional share of a yacht — capped at four shares per boat — and each share granted time tokens, one per day of use. That cap was deliberate: four shares map onto the year, so the platform could never promise more time than a boat physically has. The deeper question was what happens when you don’tuse yours, and the answer became the product: tokens were tradable, so you could spend them on sailing time or sell them back when you weren’t.

As CTO I made the architecture decisions that turned the idea into something real. Ownership and time tokens were issued on-chain to each buyer, and a booking engine connected that ownership to a real boat, a real date, a real event. The hard part wasn’t the blockchain — it was the seam between owning something digitally and using it in the physical world, and making that feel like one seamless product.

We were a four-person founding team; I led technical strategy and product engineering from concept to MVP and directed two freelance engineers. We grew a 500+ member community and earned our first real revenue from curated sailing trips and yacht-hosted events run through the platform. We wound the company down in 2025 — but I came away having taken a genuinely novel product from a human need all the way to people using it. The technology was never the point.