Personal projects, client work and hackathon builds, spanning full-stack web apps, data pipelines and machine learning.
A full-stack web app for tracking job applications in one place. It uses the Gmail API to read and categorise application emails automatically, and a FastAPI backend handles data processing and AI features via OpenAI and Ollama. Authentication and the database run on Supabase.
A research project evaluating Sentinel-2 multi-spectral imagery as a tool for tracking service delivery in South African metros. I automated bulk downloading and preprocessing of time-series imagery, then trained and validated a Random Forest classifier. The research paper is in progress.
A relational database that centralises 20+ years of historical and real-time stock market data. It includes a migration script for two decades of financial records and a daily ETL pipeline that pulls from financial APIs, replacing the company's manual data lookups.
A proof-of-concept chatbot that turns business requirements into ERP dashboard configurations. It uses the OpenAI API and manages conversational state in a Flask backend, mapping answers to modular React dashboard widgets.
Co-designed a hiring solution that replaces static CVs with cognitive-testing mini-games, using behavioural data to identify high-potential candidates. It won first place and a £200 prize.
My personal portfolio, built with AI assistance (Claude Code) in plain HTML, CSS and JavaScript and hosted on GitHub Pages. I designed the structure and content, and it's responsive with no build step.