We run a search engine for stock photography. You describe the picture you want in a sentence, in whichever language you think in, and it comes back with images that match the description rather than the caption somebody typed under it.

Underneath, every image is embedded once and searched twice: a vision model puts pictures and sentences in the same space, and a multilingual text model handles the wording. The two result sets are merged with reciprocal rank fusion. That combination is why "a quiet street in the rain" works and why the same query works in French without a translation step.

Most of our work lives inside the product, but the parts that are useful on their own are open source and sit in this organisation.

The client libraries are the obvious starting point if you want to search the index from your own code. Alongside them we publish the measurement work, since the numbers behind a vector search stack are the part nobody shares: index tuning results, cost per million images on rented GPUs, and comparisons against the other image APIs people evaluate us against. Those repositories carry the scripts that produced the numbers, so you can disagree with them precisely.

This repository also holds profilelint, which checks the page you are reading against what GitHub actually renders. It exists because the first version of this page had a centred header whose tagline rendered as literal asterisks, and nobody told us for a week. It carries a measured record of what GitHub's HTML filter does to every tag and attribute, taken from GitHub's own renderer rather than copied from a list, and a weekly job that notices when that changes.

Issues and pull requests are welcome on any repository. If something we published does not reproduce, that is a bug and we would like to hear about it, preferably with the command you ran.

Commercial questions go to hello@pexafy.com. Everything technical is better in public, on the repository it concerns.