Should I use Jev for resume screening?
Verdict for the sample idea
Run it on
Jev
A bounded judgment over text, made many times. A System One model fits.
Sample idea
For each job application, check the resume against the requirements of the role and score how well the candidate fits, so that recruiters read the strongest applications first. About 400 applications per opening.
Explanation
In technical terms this fits. The input is text, the output is a level, each application needs a small number of judgments, and recruiters cannot read 400 resumes with care. In every other sense it is the most sensitive task on this list. A wrong answer costs a person a job, and that person never learns why.
So shape it as evidence for a human, never as a decision. Do not ask "is this a good candidate?". That question hides ten judgments and all of your biases. Ask atomic questions that map to written requirements. Does the resume show three or more years of professional iOS work? Does it mention experience with payment systems? Is there evidence of team leadership? Each answer is a probability that a recruiter can check against the resume in seconds.
Combine the answers in code, with weights that your hiring team wrote down and can defend. Sort by the result. Do not hide anybody. The order changes, and the pile stays complete.
Remove the name, photo, age, address and nationality in code before the call. Jev cannot be influenced by what it never receives.
The common trap is the threshold. The moment code rejects everyone under 0.4, you have automated hiring decisions. Low confidence here means "a person reads this one first", not "skip".
The second trap is years and dates. "2019 to 2023" is four years to you and a string to Jev. Compute durations in code and send them as facts.
Related tasks
Verdict produced by Jev on 2026-09-19, rubric version 2. Text written by a person. How the tool works: the report.