Skip to content
All work

Case study · 2025 · Client project for Zaryah Angel Network, at Antier Solutions

EdgeUp

An AI-powered learning platform for institutions preparing students for UPSC, state and other competitive exams. I built the backend of its computer-based test (CBT) platform, including AI identity checks.

ID photo · at sign-uplive capture · before the paperCBT unlockedsame person ✓
Role
Sole backend developer
Stack
NestJSPythonAI / face verificationREST APIs
Website (opens in a new tab)

Results

01 The problem

EdgeUp unifies students, teachers, parents and management on one platform (AI agent, LMS and ERP). Its computer-based tests needed to be trustworthy: the person sitting the paper must be the person who registered.

02 What I built

  • Built the backend of the CBT platform for quizzes and exam papers.
  • Built an AI proctoring model in Python that captures the candidate's photo before the paper starts.
  • The model compares that photo with the Aadhaar photo given at sign-up, and only the verified candidate can begin.
  • Reused my common LMS backend architecture from EduBlock Pro and LERN360, without the blockchain layer, since EdgeUp is not a Web3 product.

03 Architecture

  1. 01 · Exam flow

    • Sign-up with Aadhaar photo
    • Photo capture before start
    • CBT papers and quizzes
  2. 02 · Backend

    • NestJS APIs
    • Exam and attempt management
  3. 03 · AI service

    • Python face verification
    • Proctoring checks

04 Hard parts

Matching faces across years

Aadhaar photos are often old. In testing, the model correctly matched me against my own Aadhaar photo, which was taken when I was in class 8 or 9.