Interactive product demo
A shorter list.
A clearer reason.
A good match should come with an explanation. Choose a sample résumé, explore the roles, and see exactly which skills connect.
02 / Explore the possibilities
Find your next direction.
Preparing sample matches…
Under the surface
A score you can inspect.
Useful rankings start with visible evidence. Open any role to see the skills found in both texts, the gaps worth reviewing, and the signals behind the score.
What actually runs here?
An unmodified copy of the extension’s lexical matcher at commit 347a854. It recognizes skill aliases, gives extra weight to stated requirements, and credits related skills when supported by specific evidence.
The optional ONNX semantic model, real listings, résumé PDF parsing, and eligibility checks are available in the extension and are outside this demo.
How should I read the score?
The score blends weighted hard-skill coverage with repeated keywords and phrases. For postings with more than three recognized hard skills, the split is 75% skills and 25% keywords. With one to three skills it is 65% / 35%; with none, only keywords contribute.
Required skills receive 1.5× weight and preferred skills 0.7×. Inferred skills receive 70% credit. Skill coverage saturates at 65%; keyword coverage has its own saturation rule. Short postings are pulled toward a neutral score. A high score is not complete coverage, verified competence, eligibility, or a probability of receiving an offer.
What happens to my text and shortlist?
Scoring happens in this tab. This demo makes no API requests, loads no analytics, stores no résumé text, and sends no text to a model provider. Reloading clears your changes and shortlist. CSV exports contain fictional roles and matching evidence; they do not include the résumé body.
All fixture content is original and fictional. Source code and license are included with the demo: matcher provenance and AGPL-3.0 license.
Bring the workflow to your search.
The extension works with your NUWorks session and adds optional on-device semantic matching.