Jev vs HR screening
HR screening is job criteria, structured interviews, and a human offer. Jev can apply a written rubric to a resume text you extracted. It is not an ATS and not a fair-hiring guarantee.
Unofficial. Employment law varies. We do not claim reduced bias. Official Jev: no LoRA on your candidates. docs.typesafe.ai.
Comparison scope
Background checks and offers stay vendors/humans. Optional: Score “mentions required license” after a parser extracts text — with audit and human override.
Criteria that decide the architecture
| Axis | Jev (System One) | HR screening |
|---|---|---|
| Offer / reject | No (policy) | Wins |
| Rubric on extracted text | Score per dimension | Recruiter |
| Protected-class inference | Do not ask | Do not ask |
| Interview scheduling | ATS | ATS |
Decision quality and control
Models inherit data and instruction bias. If you use Jev at all, log criteria, sample for adverse impact, and keep a human in the loop. This is not legal advice.
Integration trade-offs
Parser → text without photos → atomic Scores you can defend → recruiter UI. Never auto-reject on a single Noul.
TypeSafe’s public models page lists jev-1.13 at $0.042 per million input tokens with output tokens free — a vendor claim, not a jev.pro measurement. Confirm on the models page before you bid.
When each approach fits
Prefer Jev when
- You already have a written, job-related rubric and legal review
Prefer HR screening when
- Hiring decision, compensation, or background check
ATS stores; Jev optionally pre-scores text; humans hire.
Rubric scores, not a ranked slate
This is not a hiring product. ATS rank, structured interviews, and the offer stay humans and vendors. Jev may apply a written, job-related rubric to resume text you extracted — with legal review, audit, and override. Sketch: parser strips photos → atomic Scores you can defend (“mentions required license Y”) on pinned jev-1.13.0 → recruiter UI shows dimensions, not a 1…n slate → human ranks. Never auto-reject on a single Noul. Never ask a Choice among people.
$0.042 / M input is a vendor claim, not a cost-per-hire. We do not claim reduced bias. Models inherit instruction and data bias. Jaggedness: do not ask Jev to compute years of experience from date ranges — parse dates in code. No video input. Official Jev: no LoRA on your candidates. This is not employment or EEO advice.
How a “fair AI screen” becomes an adverse-impact machine
- Choice among candidates. Politics plus bias plus an unlistable class. Score dimensions; humans rank.
- Auto-reject. One schema-safe Noul is not a lawful rejection. Keep a human in the loop.
- Protected-class inference. Do not ask. Do not put photos in
state. - Generated interview questions as the product. Jev does not generate. Your ATS or an LLM under policy might — that is a different risk register.
- Rubric drift without sampling. If you use Jev at all, log criteria SHA, sample for adverse impact, and keep counsel in the review. Still not a fairness guarantee.
When not to use Jev here
Do not use Jev for offer, reject, compensation, background check, or video. Do not use it without a written job-related rubric and legal review. Prefer ATS + humans when volume is low or when you cannot defend the questions. If you cannot sample for adverse impact, do not add this hop.
What this page does not claim
-
Not an EEO product.
-
No quality-of-hire claims.
-
Schema-safe output is not the same as factual correctness.
-
No independent bake-off numbers live here.
FAQ
Can Jev rank candidates 1…n? A Choice among people is a terrible pattern (politics + bias). Score dimensions; humans rank.
Video interviews? No video input. Don’t.
Is Jev an EEO or “bias-free hiring” product? No. This site will not claim that. Counsel and your ATS own hiring.
Disclaimer
This is an independent unofficial site and is not affiliated with TypeSafe AI; official documentation is available at https://docs.typesafe.ai. Never treat jev.pro as TypeSafe official documentation. We do not sell, issue, or proxy API keys.
Hub: Comparisons. Siblings: vs onboarding checklists, questions for legal, governance guide. Canonical: https://docs.typesafe.ai.