2 of 9

THE PROBLEM

The July 2024 uprising wasn't just about the quota.�It was about a system where fraud was undetectable

and

merit was unverifiable.

Students couldn't prove the system was rigged — because there was no way to check. No transparency. No audit trail. No identity layer anyone could trust.

3 of 9

WHAT WAS BROKEN

Three Gaps Nobody Could Close

📋

Quota fraud — Seats went to people who could fabricate documents. No biometric check existed to verify eligibility.

🪪

Impersonation at exams — ID cards can be swapped. A paper token proves nothing about who is actually sitting the exam.

🔁

Duplicate NID submissions — The same national ID number could appear on multiple applications with no system to catch it.

4 of 9

THE SOLUTION

The fingerprint is the credential.

Identity links every government job application to an AI-analyzed fingerprint — from the moment of registration to the exam hall door. No card. No token. No paper.

5 of 9

HOW IT WORKS

Three Steps

Applicant submits NID, quota category, post, and a fingerprint photo. Gemini AI extracts a unique biometric profile. System issues a BDGOV tracking ID.

Verify at the exam hall

Invigilator scans the candidate's finger. Live scan is compared against the stored profile in seconds. Confidence score returned instantly.

Audit

Every verification attempt is logged with timestamp and confidence score. Administrators see quota fill rates — never personal data.

6 of 9

PRIVACY BY DESIGN

Admins see biometrics.

Never

identity.

VISIBLE TO ADMIN

→ Post applied for

→ Ministry

→ Quota category

→ FP pattern type

→ Verification count

→ Tracking ID

🔒

HIDDEN — ALWAYS

→ Full name

→ NID number

→ Father / mother name

→ Mobile & email

→ Home address

→ Date of birth

Enforced at the MongoDB query layer — .select('-personalInfo') — not just the UI

7 of 9

IMPACT

Answering the Uprising's Demands

WHAT WAS DEMANDED	HOW IDENTITY ADDRESSES IT
End quota fraud	Each seat tied to a non-duplicable biometric
Transparent hiring	Quota fill rates visible in the admin dashboard
Stop impersonation	Live fingerprint match at exam entry
Protect applicant privacy	Personal data never exposed to operators
Accountable institutions	Full audit log of every verification

8 of 9

TECH STACK

Built to Scale

⚡

Frontend

Vue 3, Vite, Pinia, Vue Router — fast SPA with bilingual Bengali/English UI

🖥️

Backend

Node.js 18, Express, Mongoose — REST API with privacy-enforced projections

🧠

Google Gemini Vision — extracts ridge pattern, minutiae, unique markers from fingerprint photos

🗄️

Database

MongoDB Atlas — separate sub-documents for identity and biometric data

🔍

Enhancement

Client-side Canvas pipeline — grayscale, histogram stretch, unsharp mask before AI submission

🚀

Deployment

Vercel multi-service — frontend + backend deployed from a single repo

1 of 9