← Software / Case study 01 Live in production

EduVerify

Attendance that can't be faked. Students must pass two independent checks, face and location, before they're marked present.

RoleSolo: research, design, build & deploy
TypeFull-stack web app, mobile-first
StackFastAPI · DeepFace · Firebase
StatusShipped · 2026
Lecturer course detail screen with a start attendance session button
Student dashboard showing a live session with a sign-in countdown
Face verification screen with an oval guide
The problem

Signing in for a friend takes two seconds.

Paper registers get passed around. Most tap-to-sign-in apps only prove that someone pressed a button, not that the right student is actually in the room. Lecturers lose class time and still can't trust the numbers.

The approach

Prove who and prove where.

EduVerify asks for two independent proofs before it marks anyone present. A face match with liveness detection confirms identity, and a GPS geofence confirms the student is physically inside the classroom.

If either check fails, the attendance doesn't count.

How it works

Three steps,
under a minute

What a student sees when a lecturer starts a live session.

01 · Location

Am I in the room?

The app reads the student's GPS position and checks the Haversine distance to the classroom's geofence.

Getting your location screen
02 · Identity

Is it really me?

A selfie runs through Facenet512 and MTCNN via DeepFace, with liveness detection to block photos of photos.

Verify identity face scan screen
03 · Confirmed

Marked present.

The record is written to Firestore in real time, and the lecturer's session updates instantly.

Attendance completed confirmation
Two sides, one system

Built for lecturers
and students

Lecturer: my courses list
Lecturer dashboard
Lecturer: classrooms with GPS coordinates and radius
Lecturers

Set up once, run every class

Create courses with join passwords, register classrooms by GPS coordinates and radius, start timed sessions and download reports.

Students

Sign in with a live countdown

Join a course with a password, get a live sign-in window when class starts, and track your attendance history.

Under the hood

The stack

Designed and built solo with an AI-assisted workflow, from research and planning through to deployment.

FrontendWeb appMobile-first, deployed on Vercel
BackendFastAPI + PythonML pipeline on Railway
Face MLDeepFaceFacenet512 · MTCNN · liveness
Data & authFirebaseFirestore + Auth, real time
LocationGeofencingHaversine distance checks
Roles2 apps in 1Lecturer and student flows
WorkflowAI-assistedClaude, ChatGPT, Cursor
StatusLiveOpen the app ↗