Screen thousands of CVs in minutes — bilingual and job-relevant in minutes
Nukhba's screening engine parses CVs in Arabic and English, extracts skills and experience, and scores each candidate against the requisition's rubric — with a defensible explanation for every result.
Everything you need to run this end-to-end
- Bilingual Arabic + English parsing and matching
- Explainable scoring — every match has evidence
- Skills gap and Emiratisation flags
- Bias-mitigation controls and human-review overrides
The single biggest waste in modern recruitment is unread CVs. On a busy req, 40–60% of applications never get a serious read; on a high-volume req, that number is closer to 90%. Nukhba's screening engine exists to solve exactly this problem — reading every CV with the rigour of a senior recruiter, in seconds, in Arabic and English, and returning a score that a human can defend.
The engine parses CVs into structured candidate profiles — skills, projects, tenure per role, education, certifications, language proficiency, location, work-authorisation signals — and scores each against the requisition's rubric. Bilingual and mixed-script CVs (a common reality in the UAE) are handled natively, not as an English-language afterthought.
Crucially, every score is explainable. The recruiter sees which lines of the CV supported which criterion, which required skills were missing, and which soft signals — job hopping, over-qualification, Emiratisation eligibility, salary expectation, notice period — the model flagged for human confirmation. Nothing is a black box, and nothing is auto-rejected without a recruiter's sign-off.
What's actually in the box
Bilingual, mixed-script parsing
PDF, DOCX, image and scanned CVs in Arabic, English or both are parsed to the same structured profile — with entity confidence scores.
Rubric-driven scoring
Required skills, preferred skills, experience bands, education, certifications, location and language weighted per requisition — recruiter-tunable in one click.
Evidence for every score
Each criterion cites the exact CV lines that supported it and the gaps that lowered it. Auditors and hiring managers can inspect the reasoning.
Bias filters and adverse-impact monitoring
Protected-category signals are removed before scoring where not job-relevant, and pass-through rates are monitored across demographic axes on every req.
Duplicate and returning-candidate detection
The engine recognises previous applicants across your history, links their profiles, and surfaces past evaluations and interview transcripts.
Emiratisation and eligibility flags
Configurable flags — Emiratisation, degree equivalence, work permit status, GCC-region tenure — that recruiters can act on without leaving the shortlist.
A government entity processes 1,200 CVs by lunch
A new call for Grade 8 analysts landed 1,200 CVs overnight. By 12:00 the following day, the recruitment team had a ranked shortlist of 42 candidates, each with evidence attached and Emiratisation-eligible candidates flagged for priority review. Previously, the team would have taken three weeks to reach this point — and would have left half the CVs unread.
Three steps, always in this order
Parse
Bilingual OCR + entity extraction turns any PDF, DOCX or image CV into a structured candidate profile.
Match
The candidate is scored against the JD rubric — required skills, preferred skills, experience bands, certifications, language, location.
Explain
Each score cites the CV evidence. Recruiters can accept, override, or send for a second opinion — every action captured.
We used to leave 40% of applications unreviewed. Now every CV gets a defensible read — and our shortlists are demonstrably better.
What buyers ask most
Ready to modernise your hiring?
Book a 30-minute walkthrough with a UAE recruitment specialist.