Things I've
built.
A selection of technical projects: coursework, personal curiosity, and things that started as a question I couldn't leave alone.
Applying machine learning methods to real estate market data. The goal is to surface patterns that traditional valuation methods overlook: spatial dependencies, temporal drift, and behavioural signals embedded in transaction histories. Under active development with academic supervision.
Built a fully playable version of Columns entirely in MIPS assembly. Implemented match detection, combo-based scoring, cascading gravity mechanics, real-time MIDI music, ghost piece preview, and difficulty scaling.
Multi-process Monte Carlo engine in C: up to 16 workers, Unix pipe IPC, π estimation and bootstrap-based statistical inference.
Interactive D3.js platform exploring 120+ years of Olympic competition: medal towers, venue maps, sport-specific metaphors, and narrative-driven design.
25 years of crude oil data: benchmark spreads, retail price pass-through, and 30+ annotated geopolitical events across WTI, Brent, WCS, Urals, and OPEC.
Relational betting platform with live data ingestion, probabilistic odds modelling, and a house-edge adjustment engine drawing from market making concepts.
Weighted collaboration graph from Spotify data: centrality metrics show eigenvector centrality beats raw collaboration count as an influence signal.
Every great developer you know got there by solving problems— Patrick McKenzie
they were unqualified to solve until they actually did it.