Explore my recent work and contributions
1st Place (Solo) at NexusLA DevDays — a full-stack ML decision-support platform that ingests thousands of Louisiana education datasets to help policymakers optimize K-12 investments across 64 parishes with AI-powered insights, policy simulation, and budget optimization.
NestScope detects birds and nests in aerial and drone imagery using lightweight YOLO-style models. Open-source NestScope Public and NestScope Experts repos, plus a CC BY 4.0 dataset on Hugging Face.
SmartPotato turns everyday PCs into a smart local recorder with natural-language search over your footage. Open source on GitHub; related dataset on Hugging Face.
An AI-powered guessing game API server inspired by Akinator, where users think of an item and the system asks yes/no questions to identify it through machine learning. You can play the game at https://paokinator.olisemeka.dev/.
PragBase is an embeddable widget that analyzes your knowledge base, and provides intelligent, structured answers. Built for businesses to create AI chatbots from their own knowledge bases.
An intelligent search engine that helps users find relevant information using natural language queries, making content easily discoverable.
Developed a sophisticated sequential fine-tuning pipeline to create a specialized medical AI model focused on dialysis care. This project demonstrates advanced machine learning techniques, domain adaptation, and responsible AI development practices using Google's Gemma 2B model.
INFERNO is a real-time fire, smoke, and motion detection system designed for edge devices. Trained on 160,000 labeled images and optimized for Raspberry Pi, it offers robust performance and rapid alerts without the need for expensive hardware or cloud subscriptions.
Find, save, and organize recipes with Meal Craft. Get meal recommendations from tags and an AI-powered Chef Steve, plus manage your own recipe collections.
End-to-end EEG data analysis pipeline: from preprocessing and statistical feature extraction to unsupervised clustering, metric-driven evaluation, and comprehensive visualization. Demonstrates robust data mining and ML engineering techniques using real neural signal datasets.
A comprehensive analysis of EEG classification using artificial neural networks compared to classical machine learning models, showcasing advanced model engineering, automated preprocessing, and robust evaluation pipelines.