Projects

Driver Drowsiness Detection System: AI Bootcamp Project

As part of my work at RAIN, I developed a real-time driver drowsiness detection system using computer vision techniques to monitor eye and mouth aspect ratios with OpenCV. The system triggers an audio alert via pygame when signs of drowsiness exceed a threshold, helping prevent fatigue-related accidents among overnight drivers transporting goods across cities in Nigeria. This Python-based solution demonstrates practical AI application with potential for deployment on edge devices. Currently pursuing funding from the Industrial Research Assistance Program (IRAP) for next-stage development.

Code | Demo | Presentation

AI-Powered Stock Price Forecaster — Independent Project

I built a web-based application that leverages Facebook Prophet and historical stock data to forecast price trends for companies like Apple, Amazon, Microsoft, and Google. Using yfinance for real-time data and pre-trained models stored as .pkl files, the app provides forward-looking insights such as projected highs/lows, best trading days, trend percentages, and investment recommendations. The system is deployed with Flask and a simple HTML interface, enabling users to select stock tickers, exchanges, and forecast durations. This project showcases the use of machine learning in financial markets to support smarter investment decisions.
Code

String: Cultivator’s 24 Hour Startup Competition Winner

My team and I built a social media application tailored specifically to students in STEM (Science, Technology, Engineering, and Mathematics) fields. The platform, named String, is designed to foster collaboration, resource sharing, and community engagement among STEM students across various disciplines. One of the standout features of String is its content-based recommendation system, which suggests posts, groups, and events based on individual user interests and activity. This personalized experience helps boost user engagement and makes String more relevant to its niche audience. We focused heavily on both functionality and user experience, ensuring the app felt intuitive and responsive across devices. Our work was recognized during the competitive showcase, where we won the ‘Best Tech’ award. This project without a doubt deepened my experience in full-stack development, teamwork, and building systems that serve real user needs.


LinkedIn post | String

CalenDue: 1st Place, Co.Hack 2025 by Co.Labs

CalenDue is an AI productivity tool that automatically parses course syllabi into structured events across Google Calendar, Apple Calendar, and Canvas, helping students effortlessly organize their semester schedules.

I used Claude Sonnet 4 for text extraction and Clerk for user authentication. I also enhanced the file uploader to support multiple syllabi uploads, implemented a “handle conflict” feature for overlapping events, added color-coded themes for priority differentiation, and integrated a Supabase Prisma database to persist user data across sessions and devices.

CalenDue won first place at the Co.Labs Co.Hack Hackathon, where it was recognized for its practical impact on student productivity.


LinkedIn post | CalenDue