Things I have built.

MMA Fight Advisor is an analytics platform for mixed martial arts, currently in stealth, that I co-founded and lead engineering for with David Mitchell (YouTube (opens in a new tab) ↗ · LinkedIn (opens in a new tab) ↗). At its core is a predictive model built in three layers: an LLM pipeline that reads the fight-week news flow, a highly efficient signal-extraction stage that turns that flow into structured features, and machine-learning models that produce per-fight predictions with calibrated confidence. Alongside it runs a topic classifier we built to sort MMA news into a classification taxonomy, so the feed the model reads is the same feed subscribers browse. Every prediction is locked and time-stamped before the event and scored afterward, so the model's record is public and cannot be edited retroactively. The platform also includes an automated news service, an odds scanner across markets, and a leaderboard that scores pundits and models on the same footing.

The Sickle Cell Knowledge and Information Network is a 501(c)(3) I founded in 2024 with one mission: make useful, reliable information about sickle cell disease universally accessible. Its first product, SickleCellPedia, is a retrieval-grounded AI agent, available on the web and on WhatsApp, that answers patient and caregiver questions from medical textbooks, peer-reviewed literature and credible clinical sources, in plain language, and declines when it is not confident. Ninety percent of people with the disease live in the Global South, where the knowledge gap is measured in decades of life expectancy; that is the gap SCKIN exists to close. We benchmark the agent against frontier models and share the results with the field at professional conferences, including ASCAT (opens in a new tab) ↗, SCDAA (opens in a new tab) ↗ and Warrior Con (opens in a new tab) ↗. I am founder and executive director.

Brother's Bank (working name ZapPay) was a QR-code mobile payments app for Nigeria, co-founded with my brother Karim while I was at Cornell. Half the world's adults had no bank account, most had a phone, and in Nigeria cash meant theft risk, unreliable ATMs and expensive interbank transfers. We ran thirty customer interviews in Nigeria, ninety percent of whom said they would use a QR-based app, built a working prototype, and designed an agent network so unbanked users could convert cash to balance.

Demeter was a digital property-title registry for Niger, again with Karim. Niger has no digital land database, no searchable owner identifier and no geolocation of plots, so titles cannot be trusted, banks will not accept land as collateral, and property tax raises 0.1% of GDP against 3–4% in France or the United States. Demeter proposed a GPS-located land registry, a biometric owner registry and a title marketplace, with the government receiving the system free in exchange for the right to monetise data and transactions. We commissioned stakeholder interviews in Niamey and obtained a government licence to pilot the product.