About this project

Most resume builders are funnels. The document is free to make and expensive to download, templates are paywalled, and the "ATS score" is a lead magnet that reveals just enough to make you pay for the rest. This project is the opposite arrangement: everything is free, the source is public, and the scoring rules are readable text files rather than a proprietary black box.

Why no AI API

An AI-powered version would be easier to build and worse in three specific ways. It would need a backend, which means your resume leaves your device. It would cost money per request, which means a paywall eventually. And it would be non-deterministic — the same resume would score differently on Tuesday, and no user could ever be told why a suggestion was made.

Deterministic rules give up some nuance and gain reviewability. Every finding traces to a rule somebody wrote and somebody else can argue with. That is a better trade for a tool people rely on during a job search.

How the offline matching works

A hand-curated skill taxonomy with an alias table — so a posting asking for "K8s" matches a resume saying "Kubernetes" — plus stop-word filtering tuned specifically for job-description boilerplate. Without that second part, every posting's top keyword is "experience" and the match report is useless.

How to make it better

The most valuable thing you can send is a lexicon gap: a skill alias we are missing, a weak verb pattern, a metric type that does not fit your profession. Each one is a small change with an immediate, testable effect on everybody scoring a resume like yours.

Disagreements about the rules are just as welcome. Every scoring rule is meant to be defensible in a single sentence — if one is not, it should change, and the only way we find out is when somebody says so. The contact page reaches us directly.