← All projects
Hiring decision support through CV screening
AdvancedAI and programming

Hiring decision support through CV screening

The user uploads a job description and dozens of CVs; the system reads and ranks them by how closely they actually match the requirements, based on meaning rather than keyword matching, and explains for each candidate what they have, what they lack, and why they're ranked there.

The problem

An HR employee reviews hundreds of applications by hand and gets tired, reading the first twenty carefully and the rest in a hurry. Traditional screening systems match words literally, so a candidate who wrote “interface development” is rejected for not writing “Frontend”.

Project image: Hiring decision support through CV screening

How it works

  1. ExtractionReading the CV text from a PDF, with OCR when it's a scan.
  2. CleaningNormalizing terms and separating experience, skills, and education.
  3. RepresentationConverting the CV and job description into a numeric representation that carries meaning.
  4. RankingCalculating similarity, weighting years of experience, and outputting the best candidates with an explanation.

Project features

  • Ranking by meaning, not wording, so “interface development” matches Frontend.
  • It shows each candidate's matching and missing skills and why they're ranked there.
  • It works on Arabic, English, and mixed texts — the reality of most Jordanian CVs.
  • An analytics dashboard showing score distribution, the most available skills, and market gaps.

Parts

All the parts you need in one place. Remove what you already have, and swap parts to fit your budget.

Number of parts
Total

Prices are estimates and may change. Parts marked “other supplier” are ordered outside our main store.

Expected challenges

These are the main obstacles to expect during the build, laid out clearly so you know them before you start.

  • CVs are irregular, with tables, columns, and images; extraction is a big part of the project.
  • Model bias is a real risk: it may favor one group's phrasing over another without professional justification.
  • Evaluating quality requires a manually labeled reference set, which is harder than building the model.
  • Applicants' data is personal, so storage and deletion policies must be well thought out.

The consultation gives you a solution for each one.

Students' experiences

Be the first to share your experience

Add a comment on the idea: +2 points. Or share your experience with photos of your build: +10 points. Every 50 points become a JD 5 discount voucher on parts orders.

Add your experience

Log in to add your experience

Need solutions to the challenges and a complete technical path?

In the consultation we give you a solution for each challenge above, plus a practical build plan from the first step to the last: technologies, libraries, wiring diagram, timeline, and tips for the defense.

  • Consultation session
  • PDF with solutions and a plan
  • 20 د.أ فقط
Book a consultation on WhatsApp →

Share the project

Project card

Hiring decision support through CV screening card
Message us