Products
Designed, built and tested by me, outside my day job. Each card says plainly how far along it is, so you know what you are getting.
Scores pay-on-delivery and transfer transactions 0–1000 for fraud risk, with reasons, so a person can review the risky ones.
Personal open-source product. Trained and evaluated on synthetic data.
The FXLedger desktop journal (Windows, v1.1.0) is shipped. The multi-asset Lolo's Ledger edition is in development.
A trading journal with a risk calculator and an educational performance coach. It reviews your own trades and never gives buy or sell signals.
Personal product. The FXLedger desktop code is public; the multi-asset edition is being built in a private repository.
Ask business data questions in plain English. It runs read-only SQL over a governed semantic layer, flags unusual KPIs and writes a weekly summary.
Personal open-source product. Built and evaluated on a synthetic Nigerian retail dataset.
A default-risk scorecard with plain reasons and an approval-policy simulator for lenders. Built on public Prosper loan data.
Research project in my portfolio monorepo, not a deployed lending system.
Churn risk, customer segments and lifetime value, with a weekly call list of who to contact first. Built on a public customer dataset.
Research project in my portfolio monorepo, not a deployed system.
A naira-first expense app with bank CSV import, auto-categorisation that learns from your corrections, receipt-text parsing, anomaly flags and a spending forecast.
Started from a course starter project, rebuilt and extended by me. Personal open-source project, evaluated on synthetic data.
Ask for a demo on WhatsApp. I walk you through it on your kind of data.
A short, fixed-price pilot shows whether it pays off before you commit.
I set it up, train your team and keep it running on a monthly plan.