Work
Systems we designed, built and still run.
The clearest way to judge an engineering firm is to look at what it has actually shipped. Two production platforms Yodly built end to end — not prototypes, and not slideware.
Project · 01
Wylo
A platform for fitness studios and their members. Training programmes and exercise libraries, meal planning built around what people actually cook, memberships, gift cards and payments — with a web application and native mobile clients on top of a shared backend.
- The problem
- Studios and coaches running programming, nutrition and billing across three spreadsheets and a messaging app, with no single view of a member and no reliable way to take money.
- What we built
- The whole system — backend, web application, native mobile clients, the payment flow and the infrastructure underneath. Designed, delivered and released as one piece of work rather than assembled from separate contracts.
- The hard part
- Memberships and payments, which is where platforms like this usually break — proration, gift cards, failed renewals, refunds. That logic had to be right from the first paying customer, and behave identically on the web app and both mobile clients, which is why they sit on one backend rather than three drifting implementations.
Project · 02
Sterodata
A data platform for collecting, structuring and querying large volumes of operational data. The hard part of most data work is not storage — it is turning what you already have into something a person or a service can ask questions of. That is what Sterodata is for.
- The problem
- Data scattered across sources, inconsistently shaped, and at a volume where handling it by hand stopped being possible some time ago.
- What we built
- The full path from raw source to answered question — ingestion, enrichment and the store they feed — sized so that collecting more is never the thing that has to stop.
- The hard part
- Cost per record. Ingesting at volume is easy to make work and easy to make ruinously expensive; the design had to keep the marginal cost of collecting more low enough that scale never becomes the argument for collecting less.
Why it matters
We carry our own pager
Firms that have never operated anything tend to recommend architectures that are pleasant to design and unpleasant to run. Having built these systems and stayed responsible for them means every recommendation we make is one we have had to live with — the deployment story, the on-call load and the monthly bill included.