Mike Oustamanolakis, PhD
← Work

Time as infrastructure

Payroll, treasury and logistics teams work across borders against calendars that disagree with each other. A quarter-end close lands on Lunar New Year and nobody notices until the payment fails.

Problem
No single authority says which days actually stop work, anywhere
Scale
241 jurisdictions · 653 subdivisions · 10,338 dated closures
Role
Architect, builder and editor
Status
Live · engine and API · reissued monthly
The timesys.pro homepage: the headline ‘Your quarter-end close hits Lunar New Year’, with a bar of counted figures beneath it.
The problem stated in nine words, then the counted scope of the answer.

The problem

Every organisation operating across borders needs to know which days are working days — and there is no authority that says so. Public holidays are set by gazettes, exchange notices, central banks and revenue authorities, each publishing in its own format, on its own schedule, for its own jurisdiction. Below the national layer, subdivisions diverge: a rule that holds in Bavaria does not hold in Berlin.

The commercial answers are date lists — flat files of days for a given year, bought annually, silently stale the moment a rule changes. A list cannot tell you why a date is a date, cannot be projected forward, and cannot be checked.

What I built

A rules engine rather than a calendar. Observances are stored as rules — 3,990 of them, distinct rules rather than per-year rows — evaluated across 36 calendar systems, so a date can be generated for any year rather than looked up in a table someone typed. Every row carries its source and a grade for how firm that source is.

One corpus, three deliveries: an API for systems, an Operating Calendar for teams, and dossiers as PDF, CSV and ICS for the people who will not integrate anything. The delivery changes; the corpus does not.

The timesys.pro tools section, showing the instruments built on the corpus.
Tools built on one corpus. Adding a delivery costs a view, not a dataset.

The decision that shaped it

The engine states its own edition and the moment it was last updated, on the front page, and says a rule holds for years we have never published. Publishing the limit of the data is what allows the rest of it to be relied on. A vendor claiming complete global coverage is claiming something no gazette-derived dataset can support.

Where a date is projected rather than gazetted, it is labelled projected. That single convention is the difference between a dataset a treasury team can put in front of an auditor and one they cannot.

The timesys.pro pricing page, showing the tiers for API, calendar and dossier access.
Three products from one corpus, priced by how the buyer wants to consume it.

What it demonstrates

Turning a scattered, authority-fragmented domain into infrastructure other systems can depend on — and being explicit about where the authority runs out. The same shape recurs in waypoints.now, where the unit is reach rather than a route.

Visit timesys.pro →