Selected work
Four live sites you can use, and six repositories you can read. Nothing here is a mockup or a tutorial follow-along.
Web & product engineering
An electronics storefront, a bookstore checkout, a resort booking flow, and a charting tool. All four are running right now.
Full-Stack Storefront
The storefront for an electronics shop in Kampala, which I built end to end. A React and TypeScript single-page app on Supabase Postgres: 100+ products with row-level security, an admin panel the non-technical owner uses to manage stock and upload images, search, and enquiry-led buying instead of checkout. Deployed on Vercel with preview builds for every branch.
Full-Stack E-commerce
A collaboration on the online storefront for an established Ugandan bookstore chain, and one where I worked across the full stack: catalogue browsing and search, customer accounts, wishlist, cart, and checkout with delivery or in-store pickup, built as a React application against a token-secured API. The platform is live and still in active development.
Time-series Visualisation
Hospitality Booking Platform
A collaboration on the booking site for a nature resort in Buwate, Wakiso: accommodation, dining, events and activities, with a date-and-guest availability search on the landing page and WhatsApp enquiry built into every screen. A React single-page application bundled with Vite, served over nginx.
Data engineering & analytics
Projects that organise, assess, analyse, or communicate data.
Power BI
A five-page Power BI report over 41,188 telemarketing contacts, showing a bank that cut 93% of its calls and lost only 20% of its deposits — a 10.8x lift in conversion. The source file carries no year, so the campaign calendar is rebuilt in Power Query from row order and the consumer price index. Authored in PBIP format, so the model and every visual stay reviewable as plain text.
Data Engineering
Scores a dataset for missing values, duplicate keys, bad formats, impossible dates, out-of-range numbers and orphaned foreign keys. Use it in the browser (the Python engine runs locally via Pyodide, so files are never uploaded) or as a CLI quality gate in pipelines.
Exploratory Analysis
Exploratory analysis of medical insurance data to examine cost drivers and demographic patterns in expenditure.
Applied machine learning
Projects demonstrating dataset preparation, modelling workflows, and domain-specific problem framing.
Computer Vision
A convolutional neural-network project that classifies chest X-ray images for pneumonia detection.
Classification
A classification workflow for identifying customers who may be at risk of leaving, supporting retention analysis.
How to review the work
The project page stays concise; source code and repository documentation remain the primary evidence.
Start with the repository description and the real-world question the project is framed around.
Review data preparation, analysis or modelling choices, dependencies, and documented assumptions.
Inspect notebooks, scripts, visualisations, and available live demos without relying on unsupported claims.
Share the users, the data behind it, and the result you want to make repeatable.