I built a deterministic, quote-verified ATS resume scanner & job aggregator so AI stops hallucinating fake skills

Описание и применение

Создан детерминированный сканер резюме с проверкой цитат, устраняющий случайные оценки и выдуманные навыки, обеспечивая стабильные и проверяемые баллы.

Задача

AI resume scanners produce random scores and hallucinate skills, causing unreliable assessments and candidate rejection.

Результат

Consistent ATS scores with 55% weight on quote‑verified hard skills, 15% measurable impact, 15% soft skills, 15% structural completeness, eliminating hallucinations and score variance.

Рабочий процесс

• Extract job description skills
• Parse resume into sentences
• Verify each hard skill by matching an exact sentence quote
• Extract measurable metrics via regex/NLP
• Check soft skills and role markers
• Validate structural completeness (contact, links, headers, word count)
• Calculate weighted ATS score from verified components

Доказательства

Random Score Generation: Because LLMs are probabilistic (temperature > 0), submitting the exact same resume twice often yields 62% first and 84% a minute later.

Источник

Skill Hallucinations: When asked to tailor a resume or cover letter, standard LLM wrappers cheerfully invent experience with technologies the candidate has never touched—leading to instant rejection during the first 5 minutes of a technical screen.

Источник

55% — Quote-Verified Hard Skills (resumeQuoteEvidence): A skill from the job description is only counted as matched if the parser extracts the exact sentence from the resume as proof.

Источник

Here is the core deterministic scoring and quote-extraction logic from our engine: ...

Источник