JD Keyword Extractor

Not every requirement is a requirement

A job posting is a wish list with a few hard filters buried in it. "Strong experience with Java" and "exposure to Kubernetes is a plus" are not the same requirement, and treating them as equivalent is why people either over-prepare for a nice-to-have or fail to notice the one thing they genuinely need.

So this tool sorts what it finds by how hard the posting asks. Must-have covers what was stated as a requirement. Preferred covers what was named as a plus, a bonus or a nice-to-have. Mentioned covers everything present without a stated level of importance - worth knowing, not worth reorganising your preparation around.

It matches meaning, and it says when it could not

The terms found in the posting are matched against our own curated skill taxonomy rather than compared as strings, so a posting asking for "Spring" and a posting asking for "Spring Boot" land in the same place. When that matching service is unavailable, the tool falls back to exact wording and labels the result as such - a thinner answer, clearly marked, rather than a silent downgrade.

There is also an unmapped list, and it is the part worth reading. Terms that stood out in the posting but do not match anything in our taxonomy are shown rather than dropped: usually a company-specific tool, an internal system, or a phrase that is not really a skill. A list that quietly discarded what it did not understand would look complete and be wrong in the flattering direction.

Free forever, and there is no AI in it

This tool costs nothing and will continue to cost nothing. It also never calls a language model - the parsing and the taxonomy matching are our own code, which is why it is fast, why the same posting always produces the same list, and why there is no consent checkbox on the page.

The recruiter's contact details in the posting are removed before anything is stored. We do not keep or surface third-party contact information, which is the one piece of a job description that is somebody else's personal data rather than yours.

What this tool cannot do

It cannot tell you whether you are a fit. It never sees your resume, so it has nothing to compare the posting against. That is the JD ↔ Resume Match tool, which takes both and tells you which requirements you evidence and which you do not.

It cannot tell you how a company's screening filters are configured. Nobody outside that company can. This is what the posting asks for, in the posting's own order of priority - which is a different and more reliable thing than a guess about an applicant tracking system.

It will not help you stuff keywords. Cramming a posting's phrases into a resume produces a document that reads as spam to the human who opens it next, and that human decides.