Biometric Attendance System

I built a custom Arduino library for the Nano RP2040 Connect using the JM-101 fingerprint sensor to power an IoT biometric attendance system. The system supports full local functionalities and remote RESTful operations over Wi-Fi:

• GET /fingerprints — Get count of stored fingerprints
• GET /fingerprints/{id} — Check if a fingerprint exists
• POST /fingerprints/enroll — Enroll new fingerprint (JSON: {"id":1})
• PATCH /fingerprints/{id} — Update/re-enroll a fingerprint
• DELETE /fingerprints/{id} — Delete specific fingerprint
• DELETE /fingerprints — Delete all fingerprints

The library handles sensor initialization, enrollment, matching, and deletion, while the REST API allows integration with external systems for attendance tracking and management.

ClearUI Surveillance

Traditional surveillance infrastructure require huge capital investment, complex wiring, proprietary hardware, and professional installation. ClearUI eliminates these barriers with a distributed edge architecture that deploys in minutes rather than days. The system aggregates multiple camera endpoints through a unified network gateway, delivering low-latency video streams directly to any display or mobile device on the local network. Remote access is secured via WebRTC, enabling real-time monitoring from anywhere without recurring cloud subscription fees or third-party dependencies.

ESP32 + Go ML

A home automation system with an HC-05 ultrasonic sensor sending distance data to a PHP server via Wi-Fi. Machine learning performed with Go (Gorgonia) and Arduino for predictions.

LibSecAshesi

An IoT security system to prevent book theft at Ashesi University's library — detecting when books leave without authorisation and alerting staff in real-time using RFID.

Thermal Imaging

Visualises real-time heat signatures from the MLX90640 IR array via serial output and a dynamic Python heatmap. Built with pyserial and matplotlib — interfaced through an ESP32 with a custom 3D-printed housing combining a serial camera and VL53L0X depth sensor.

Pytorch ARM Workshop

To celebrate the PyTorch and ARM collaboration in 2023, I led a hands-on workshop demonstrating how deep learning models can be deployed on ARM-based hardware. Participants learned to train a neural network for rain prediction using environmental sensor data, then run live inference on edge devices—bridging the gap between cloud-based training and on-device AI.

Jarvis

I built this project to give my ARM Based Single Board Computer capabilities similar to a voice assistant but the ability to perform local actions. It listens for wake words using Porcupine, captures what I say with speech_recognition, and talks back through pyttsx3. When I ask it to do something, spaCy figure out what it means. It integrates with microcontrollers via PySerial—sending commands back and forth over a simple serial link enabling high-level voice control over to physical hardware.

RLE Compression

Built an image compression pipeline from scratch using Run-Length Encoding (RLE) on grayscale images — covering quantization, lossy vs lossless techniques, and achieving up to 60% reduction in file size at C level speed.

BlockBusters

A 2D game where you compete against an A* search algorithm. Control mode and competitive mode sessions — race to earn the most points with randomised role-play and chance elements.

Personally Identifiable Information Detection Using Most Frequent Class (MFC) and Next Most Frequent Class (NMFC)

This project teaches a computer to identify private information like names, emails, and phone numbers in text. I built simple prediction models that learned from thousands of examples, achieving 85% accuracy in identifying sensitive information.

Coolors++

A reimagination of Coolors.co — lets users anonymously share colour themes from their projects. Real-time updates via WebSockets and a REST API backend.

Gunzip Extractor

A simple, fast tool for extracting .gz files without the need for command-line tools or complex software.

The app lets users select any .gz file and instantly decompress it to its original format—whether it's a JSON, CSV, text, or other file type. No terminal commands, no installation headaches.

Built with Flutter, the app works across desktop and mobile, providing a clean drag-and-drop interface with real-time extraction status. Perfect for developers, data analysts, or anyone who regularly works with compressed files.

Qlear

A mobile platform for creating and sharing forms, polls, and event invitations, etc, without the complexity or cost of tools like Google Forms or Eventbrite.

Qlear's standout feature is Secure Form. Secure Form is a simple invite-only system where each attendee receives a unique QR code by email, which serves as their ID. Event organizers just scan the codes upon arrival and hence, no need to print attendance cards or check IDs manually.

Built with Flutter, Firebase Realtime Database, and a custom local database with intelligent processing, the app dynamically generates forms from questionnaires and allows instant sharing via link or QR code.

VariableX

A TypeScript-powered platform with a statically generated Next.js frontend. Designed for performance and scalability with modern web development best practices.

Ashesi ARM Student Club

Building the future with ARM through workshops, hands-on projects, and microcontroller innovation. Empowering students to create ARM-based solutions.