What it is
A recruiting screening project that scores one resume against a DeepSeek user-operations JD. It packages criteria templates, test requests, results, resume samples, and a vanilla-JS web workbench for HR reviewers.
How it uses Jev
Jev answers ten atomic questions in one request: five Noul evidence gates, four Score dimensions (0–3), and one Choice for background routing. Code normalizes scores, applies hard gates, flags low-confidence answers for human review, and routes by background bucket.
Primitives:choicescorenoul
Technique worth stealing
Split the judgment into atomic questions, keep criteria language fixed, and let code synthesize the total score.
Try it
Open console.typesafe.ai Playground and send requests/*.json with model jev-latest.
Evidence
Each line is one question put to Jev about the README. ≥ 0.60 reads as yes, ≤ 0.40 as no; in between Jev is not making a call.
- Jev-centricyes0.97
- Shows a System One patternyes0.97
- Handles uncertaintyyes0.90
- Measuredyes0.83
- Runnableno0.13
- Worth recommendingyes0.80
- Model replicano0.04
- Problem scopescore on a 0–2 scale1.18
- About Jevyes0.99
Signals by Jev jev-1.13.0, card written by DeepSeek V4.1 Flash from the README on 21 Sept 2026.