Data · Technology · Communication
I work with data, build with technology, and write about both in ways that non-technical people can actually use.
I'm a technologist who moves comfortably across data science, software development, cloud infrastructure, and product thinking. My background in Computer Science gives me the technical range — from building ML pipelines to implementing cryptographic systems — while my minor in English Literature and Language means I have never treated communication as an afterthought.
I've geocoded thousands of polling units to surface electoral anomalies, built air quality forecasting models for cities with barely any infrastructure, and designed visualisation dashboards that tell stories rather than just display numbers.
Outside technical work, I write. My Medium blog explores product thinking, technology decisions, and honest reflections on being in transition — because explaining complexity clearly is a skill I take as seriously as building things.
"I don't just run the analysis — I can tell you what it means, and what to do next."
"A very cleanly written report. Much of it wouldn't be out of place in a computer science journal article. The XAI analysis is particularly impressive — the level of reporting, in terms of both the way data is presented and the depth of analysis, is what I'd expect to see in a journal article."
Next-day PM2.5 forecasting across three under-monitored cities using only open data — satellite night-time lights, meteorological reanalysis, and ground sensors. Compared XGBoost against deep learning sequence models. Applied SHAP explainability to make model outputs usable by urban health decision-makers, not just data scientists.
Geocoded 4,325 polling units, manually corrected 17 erroneous location entries, and applied geospatial neighbourhood analysis to compute voting outlier scores for each party. Built interactive maps showing the top 3 anomalous units per party — making statistical findings navigable without reading a report.
Group project classifying human vs. natural wildfire causes across a 28-year US dataset. Benchmarked seven algorithms — including CatBoost, CNN, MLP, and Decision Trees — with and without SMOTE for class imbalance. I led all documentation and progress tracking, wrote the full technical report, implemented K-Means clustering, and contributed collaboratively to model development.
Data-driven flood risk prediction for Lagos using spatial and environmental features. Identifies high-risk areas to support urban infrastructure planning and disaster response prioritisation.
Analysed 645 property transaction observations using an inverse gamma model. Derived MLE and Method of Moments estimators analytically, computed Fisher Information, and built a 90% confidence interval for the rate parameter. Validated model fit by simulating 10,000 predicted mean values — confirming tight distributional alignment with observed data.
Led a four-person team building an interactive D3.js dashboard exploring genre profitability, ratings, and financial performance. Designed the layout, built the bubble chart with square-root scaling, established the Git workflow that eliminated merge conflicts, and managed scope — steering a complex visualisation from near-abandonment to delivery.
Java OOP application for managing a memorabilia auction inventory with a full Swing/AWT GUI — JList display, JOptionPane dialogs, dual Comparator sorting, and real-time CSV data persistence. Built with iterative CI/CD thinking: issues tracked per feature stage, commits staged to demonstrate incremental delivery.
A product feature audit of X (Twitter) that goes two levels deep — first analysing the platform's design decisions against Ravi Mehta's Strategy Stack and Elon Musk's Algorithm, then turning the same critical lens on the audit itself. An honest account of what it means to think like a product person when you're still learning the vocabulary.
A candid reflection on entering the product management space in 2026 — the noise, the certainty everyone else seems to carry, and what it actually feels like to be in the early stages of a transition. Draws on Marc Andreessen's recent observations about the shifting PM role.
Looking for roles where data and technology work is connected to real decisions. If that sounds like your team, drop me an email or reach out on LinkedIn.
