Presensee

Presensee is a smart internal attendance platform from specific institutions that use facial recognition, some verification, and a powerful admin dashboard.

Executive Summary & Role

Me and my friends built Presensee to secure an internal attendance system for a specific institution (sorry, the institution needs to keep some information private, so I can’t provide specific details).

This is my first non-monolithic project, it has backend with Laravel and a Flutter for the frontend, also its uses a Python and Dlib for facial recognition system.

Key Features

Face Recognition: The backend runs a Python computer vision pipeline that compares captured image from flutter using camera against data in the database using specific verification.
On-Device ML Detection: The Flutter mobile app uses Google ML Kit to detect a face locally before even sending the image to the server, saving bandwidth and preventing spoofing.
Powerful Admin Dashboard: The admin dashboard provides a comprehensive view of attendance data, and management tools.

The Problem & The Solution

The Problem:

Buddy Punching & Fraud: Traditional methods make it way too easy for employees to clock in for absent colleagues.
Hardware Bottlenecks: Physical device (e.g., fingerprint scanners) cause massive queues during peak hours, cost a lot of money, and break down constantly.
Admin Overhead: Some head of departments spend too much time managing some administrative tasks also sometimes they have miscommunication each other.

The Solution:

Biometric Smartphones: By moving attendance to the employees’ own phones, we eliminated hardware queues and maintenance costs entirely.
Strict Verification: The combination of Python-based Facial Recognition and strictly verification algorithms makes faking attendance nearly impossible.
Centralized Admin Dashboard: The admin dashboard provides a single source of truth for attendance data, reducing miscommunication and administrative overhead.

Technical Details & Architecture

This is where the tech stack gets really exciting. Here is why I decoupled the architecture:

Laravel: Acts as the central brain. It handles the database, business logic, and issues secure API tokens to the mobile app.
Python & Dlib: Laravel communicates to a Python script running face recognition and some algorithm to do some verification.
Flutter & Google ML Kit (Mobile): Android app that uses native camera integrations, Google ML Kit to detect faces locally before sending the images to the Laravel API.

why not monolithic? A mobile app requires a completely different client than a web dashboard. By decoupling them, the Flutter app and Laravel dashboard can evolve independently.

why Python for Face Recognition? PHP isn’t built for heavy math and machine learning. Python has tested computer vision libraries that process facial recognition with good accuracy.

Challenges & Learnings

Building a decoupled, AI-powered system brought a whole new set of challenges:

Bridging PHP and Python: Figuring out how to efficiently pass images from a Laravel controller to a Python execution script and parse the response without blocking the server was a massive learning curve.
Machine Learning: Leraning about the fundamentals of machine learning and how to apply them in a production environment.
[some redacted challenges and learnings]

Key Takeaways & Results

Presensee was a massive milestone for me. It forced me to step out of the standard “monolithic web app” comfort zone and dive into the world of distributed systems and mobile development.

Mastering Decoupled Architecture: I now fully understand how to build secure, scalable REST APIs that serve multiple distinct clients (web dashboards and mobile apps).
ML Integration: Successfully integrating a Python machine-learning pipeline into a standard PHP backend proved that you can combine the best tools for the job.
Mobile Dev: Learning Flutter and integrating it with device hardware (cameras, etc) gave me the confidence to build full-stack cross-platform apps.