LIMITED-TIME OFFER: Get up to 15% OFF on Cloud Billing + FREE Cloud & DevOps Consultation Claim Offer LIMITED-TIME OFFER: Get up to 15% OFF on Cloud Billing + FREE Cloud & DevOps Consultation Claim Offer LIMITED-TIME OFFER: Get up to 15% OFF on Cloud Billing + FREE Cloud & DevOps Consultation Claim Offer
DevSecCops
Home
Case Studies
HR TECH • RECRUITMENT AI • CASE STUDY

A Ranked Shortlist, With the Reasoning Attached

Reads the job description and every resume against it, and hands the recruiter a ranked shortlist with the reasoning attached, in minutes rather than days.

LLM Scoring RubricJD ExtractionParallel Reviewer AgentsRole-Based Access
HR TECH · RECRUITMENT AI
LLM Scoring Rubric
JD Extraction
Parallel Reviewer Agents
64.7sAverage time from upload to scored result (112s worst case)
$0.16AI cost per resume — ~$32 for a 200-resume role
~4/minSustained throughput, audited September 2026
The Challenge

A recruiter is judged on four profiles — everything before that is unpaid, unscalable work

A role can draw a few hundred resumes across email, Drive folders and forms. Read properly it takes days; read quickly, good people are missed. The JD is the real spec but is rarely read closely, so two recruiters weigh skills, years and domain differently and the client gets an inconsistent bar. Nothing remembers: a strong engineer who applied for another role last month is invisible, and every search starts from zero.

HR Tech AI Solutions

Want a Shortlist Scored on Your Own Job Descriptions?

We'll scope a pilot around your exact recruiting workflow — JD-locked scoring, ranked shortlists and auditable results — with clear success metrics before any production commitment.

Book a Free AI Scoping Call
What It Is

The platform does the reading

It scores every candidate on the same rubric, shows why each score came out as it did, and keeps shortlist, phone screen, interview and offer in one place. The recruiter still decides who goes to the client.

1

JD goes in

Requirements are extracted and the recruiter reviews and corrects them before the rubric locks to this role.

2

Resume text extracted

Text is pulled from PDF or document in about 0.7 seconds.

3

Six reviewers read in parallel

Skills, experience, role and domain, education, communication, availability — about 60 seconds, 98% of the total.

4

One score, broken down

0 to 100 overall, per dimension, with what the resume actually said. 75% of mandatory skills gates the shortlist.

5

Ranked shortlist

Shortlist, verify on the call, or reject — the recruiter decides.

Speed and cost — production, audited September 2026

A month of live traffic, measured end to end.

MetricValue
Upload to scored result (average)64.7s
Upload to scored result (worst case)112s
AI analysis share of total time63.4s (98%)
Text extraction0.7s
Waiting in queue0.5s
AI cost per resume~$0.16
Sustained throughput~4 resumes / minute
A 200-resume role~$32 of AI spend

Why the score can be trusted

  • The JD sets the rubric, not the resume, so everyone is measured on the same scale
  • Mandatory skills gate the shortlist (defaults: mandatory 60%, must-have 30%, preferred 10%)
  • Thin evidence routes to a human — the platform flags the skill and says what to verify on the call
  • Every score is traceable and reversible: plan, breakdown and evidence stored; recruiters can correct and override

After the shortlist

Intake from uploads, Drive or forms with deduplication; phone-screen recordings transcribed and summarised; Google Meet interviews joined, transcribed and analysed; custom client rounds (L1, L2, HR) logged with who and when; client-ready PDF, CSV or Excel exports with scores and reasoning. Candidate data is visible only to people who own or are assigned to the role, enforced on the server and covered by tests.

Status: running in production — intake, JD extraction with review, JD-locked scoring, pipeline, phone-screen and Meet analysis, exports, analytics, role-based access. Built, awaiting rollout: talent-pool search; external evidence from GitHub, Medium and LinkedIn (shown, not yet scored). In flight: faster analysis (target about 20 seconds) and round-level timing.

Why Build This With DevSecCops.ai

An engineering team that ships audited production systems, not slide decks

AWS Advanced Tier Partner

Certified cloud and AI delivery, so your hiring platform runs on infrastructure built to AWS Well-Architected standards from day one.

Pilot-first engagement

We scope, build and validate against real resumes and job descriptions before you commit to a production rollout.

Traceable, reversible scores

Every score's plan, breakdown and evidence is stored, and recruiters can correct and override it — not a black box.

Role-based access, enforced server-side

Candidate data is visible only to people who own or are assigned to the role, enforced on the server and covered by tests.

Trusted by forward-thinking teams

Cars24
Dekoder
FlexiLoans
Goodmeetings
Hypestore
Frammer
Hindustan
ShareChat
Microsoft Solutions Partner
Cars24
Dekoder
FlexiLoans
Goodmeetings
Hypestore
Frammer
Hindustan
ShareChat
Microsoft Solutions Partner
AWS CERTIFICATIONS

AWS-Certified Experts Delivering Cloud Excellence.

Solutions Architect
DevOps Engineer
Security Specialty
Cloud Practitioner
Data Engineer
Partner Badge
Partner Badge
Partner Badge
Solutions Architect
DevOps Engineer
Security Specialty
Cloud Practitioner
Data Engineer
Partner Badge
Partner Badge
Partner Badge

More HR Tech AI Solutions

HR Tech AI Solutions

Want a Hiring Platform That
Scores on Your JD?

Tell us about your recruiting volume and process — we'll help you scope an AI hiring pilot with JD-locked scoring and auditable, overridable results.

JD-locked scoring so every candidate is measured on the same scale
Traceable, reversible scores — not a black box
Audited production metrics, not projections

Send us a message

Fields marked * are required.

By submitting you agree to our Privacy Policy. We never share your data.

Talk to an Expert