OpenAI · RAG · CV Parsing · AI Ranking · HR Tech

HR Venture CV Engine: Smarter Screening, Faster Shortlists

An intelligent CV screening engine that automates candidate evaluation at scale — uploading bulk CVs, parsing skills and experience with OpenAI + RAG, scoring each candidate against a custom job description and persona, and returning ranked shortlists with explainable, evidence-backed scores.

191
CVs Processed
Single role, live run
88%
Top Match Score
Explainable AI scoring
RAG
AI Parsing Engine
OpenAI + retrieval
0
Manual Screening
Fully automated

What We Built & Why

HR Venture needed to screen hundreds of CVs for multiple executive roles simultaneously. Manual screening was taking weeks, introducing inconsistency, and making it impossible to apply role-specific evaluation criteria at scale.

We built a full-stack AI CV Engine that lets recruiters define a hiring persona, upload a job description, configure a custom scoring framework, and bulk upload CVs. The engine uses OpenAI + RAG to extract skills, experience, and achievements from every resume — then scores, ranks, and returns a shortlist with an Evidence Pack for each candidate.

Core problem solved: Screening 191 CVs manually for a CIO role took weeks. With CV Engine, 189 CVs were processed in 75 minutes — returning a ranked shortlist with explainable match scores, evidence points, and candidate fit reasoning ready for the recruiter to review instantly.

Persona-Driven Scoring

Recruiters define the ideal hiring persona — seniority, priorities, deal-breakers. Scoring adapts per role.

Bulk CV Upload

Upload hundreds of CVs in one go — PDF, DOCX, any format, all processed automatically.

Explainable Scoring

Every score backed by evidence from the CV — no black-box decisions, full recruiter transparency.

Ranked Shortlists

Top 10, 20, 50, or full ranked output — plus Excel export for downstream hiring workflows.

CV Engine — Running in Production

HR Venture CV Engine — Upload Role + CVs Batch form with Company, Job Title, JD, Persona, Scoring Framework, and CV upload
Upload Role + CVs Batch

Recruiters input the Company and Job Title, upload the Job Description (PDF/DOCX), optionally upload an Ideal Candidate Persona, define a Scoring Framework (XLSX/PDF), then bulk upload all candidate CVs in one batch.

· Job Description (PDF/DOCX) — core evaluation context
· Ideal Candidate Persona (optional) — aligns scoring
· Scoring Framework — custom weightage per role
· Candidate CVs — select files or upload entire folder
Roles Dashboard — Processing Status

The roles dashboard shows all active processing jobs — status, role, company, time taken, and progress bar. Four roles completed for Probat and Breitling — CIO (128 mins, 191 CVs), GHM (70 mins, 185 CVs).

· CIO without persona — 189/191 processed in 75:44
· GHM without persona — 184/185 processed in 70:21
· GHM — 185/187 processed in 62:52
· CIO — 188/191 processed in 128:43
HR Venture roles dashboard — 4 completed roles for Probat and Breitling with processing progress
HR Venture CV Engine results — CIO role, 191 CVs, candidate Marcin Palmer ranked #1 with 88.2% match score and Evidence Pack
Ranked Results — Evidence-Backed Shortlist

CIO role for Probat — 191 total CVs, 189 processed, 2 duplicates removed. Candidate #1 Marcin Palmer ranked with 88.2% match score and a full Evidence Pack explaining exactly why he fits.

· Filter: All, Top 10, 20, 30, 50, 60+%, 80+%
· Evidence Pack — detailed fit reasoning per candidate
· Search by name, skill, or email
· Export Excel — full ranked list for downstream review

The AI Architecture Behind the Engine

The recruiter defines context once — persona, JD, scoring weights. The engine handles everything else: parsing, analysis, scoring, and ranked output.

RAG-Grounded Scoring

RAG ensures every score is grounded in actual CV content — no hallucinations, no guesswork. Each data point traced to source text.

#1 3rd 2nd 1st
Persona Alignment

The same CV scores differently for a junior vs senior role. Persona alignment means scoring always reflects what the role actually requires.

Custom Weightage

Recruiters set the scoring framework — prioritise skills over experience, or vice versa. Every role can have a completely different evaluation model.

JD CV AI Score + Rank
Explainable Output

Every score has a reason. Recruiters receive an Evidence Pack per candidate — builds trust and enables confident, defensible hiring decisions.

The CV Engine Pipeline — Step by Step

01
Define Hiring Persona

Recruiter inputs hiring persona — seniority level, key priorities, and deal-breakers to guide how scoring is applied across all CVs.

02
Upload JD + Scoring

Job description uploaded as evaluation context. Custom scoring criteria and weightage configured for this specific role.

03
Bulk CV Upload

Multiple CVs uploaded in one go. The engine accepts varied formats and prepares all documents for AI analysis.

04
AI CV Parsing

OpenAI + RAG extracts skills, experience, education, and achievements from each resume regardless of formatting or structure.

05
JD Matching

Each extracted profile compared against the job description using the RAG pipeline to identify alignment and gaps per candidate.

06
Explainable Scoring

Each CV receives a weighted score based on evidence found in the document — with clear reasoning for every data point.

07
Candidate Ranking

Candidates ranked by score. Recruiters receive Top 10, Top 20, Top 100, or the full ranked list for review.

08
Decision Support Output

Recruiters receive a shortlist with scores and reasoning — enabling fast, confident, evidence-backed hiring decisions.

What CV Engine Delivers

191 CVs screened in 75 minutes — a process that previously took hiring teams days of manual review is completed automatically before the next morning.

Zero subjective bias — every candidate scored by the same AI model against the same JD and criteria, eliminating inconsistency across reviewers.

Ranked shortlists with Evidence Packs — recruiters start interviews knowing exactly why each candidate is a fit, reducing mis-hires significantly.

Scales to any volume — the same pipeline handled CIO and GHM roles for Probat and Breitling simultaneously, with no infrastructure changes.

Custom scoring per role — finance roles can weight financial acumen, tech roles weight engineering depth, without any code changes.

Technologies Used

OpenAI API
LLM parsing & scoring
RAG Pipeline
Retrieval-augmented gen
CV Parsing
PDF & DOCX processing
AI Agent
Autonomous scoring
Python
Backend & pipeline
Full Stack App
React + Node.js

What Was Built & Applied

OpenAI API Integration
Retrieval Augmented Generation (RAG)
AI Agent Development
CV Parsing & Document Processing
Scoring & Ranking Algorithms
Full Stack Application Development

Want to automate CV screening for your hiring team?

From bulk CV upload to RAG-powered scoring and ranked shortlists with Evidence Packs — we build AI hiring engines that process hundreds of candidates in hours, not weeks.

Book a free call →