Article
How to read this site
This site is a working portfolio and, at the same time, a small lesson in how one can be built honestly. Every number on it comes from a script that measured it: the repositories on GitHub and GitLab, read through their command-line clients, the contribution calendar, and the date of each project’s last commit. Nothing is typed twice: the pages of categories, projects and terms are derived from data, and a check refuses a stale copy.
The pictures are data visualisation drawn with D3.js from the same measurements. The galaxy places every original repository as a star, one arm per category; the pulse is the only picture that is not a snapshot: it asks the public GitHub events API from your browser and redraws the last weeks of commits, pull requests and releases as a heartbeat, every five minutes. The spiral lays a year of contributions around a circle, one ring per weekday, so busy seasons glow; the bubbles size each language by the repositories written in it; the streams show how those languages rose and faded over the years. Hover any of them to read a value, click to open the work behind it.
Who wrote what matters. A fork or a mirror is counted but never shown as the owner’s work: a repository is original only when the owner wrote at least half of its commits.
The site itself is built from front matter and data by Zola, a static site generator, styled with Flowbite Pro, and checked against WCAG 2.2 AA in both themes before it ships. Machines read the same facts as JSON-LD, as llms.txt and as site.json; social networks draw each page’s preview from its Open Graph tags. Every page ends with the concepts it builds on; the glossary defines each one and cites a source.