02 / THINKING A FEW MOVES AHEADADAPTING TO NEW CHALLENGES
3/465Optiver AI Chessathon
A chess engine with a need for speed.
465 teams from around the world entered. We topped the Swiss qualifiers, reached the 32-team London knockout, and finished third overall. Three builders, one CPU core, and about 900,000 nodes searched per second.
1ST IN SWISS QUALIFIERS · 3RD IN FINALS~40% FASTER WITH AVX2 ↗︎
03 / THREE DAYS, ONE IDEALEADING A TEAM
Cambridge HackEdHackathon Winner
An AI platform that turns lessons into revision tools.
We built a web app that turns uploaded lesson content into notes, flashcards and practice questions using AI. I led our three-person team and co-developed the platform, bringing the Next.js frontend and Django backend together in three days.
3 PEOPLE / 3 DAYS / ONE WORKING PLATFORMNEXT.JS FRONTEND / DJANGO BACKEND
04 / AI AGENT PAYMENTSDESIGNING SAFEGUARDS
Cantor8 Hackathon4th Place
A wallet that lets AI agents spend within limits you set.
Our five-person team built an AI-agent wallet with spending rules enforced by smart contracts. I implemented per-period and lifetime caps, predictable allowance resets, and controls to revoke an agent’s spending access. We won a £1,000 team prize.
PythonDamlSmart contracts
agent✳︎
▦
Big ideas. Bounded budgets.
£1,000TEAM PRIZEconcept illustration · real safeguards
Python simulations of how light bends, focuses and splits into colour.
I built models and interactive visualisations across twelve computational tasks, exploring lenses, prisms, refraction and rainbows. I used ray tracing and numerical methods to compare the physics with simulated and experimental data.
An eight-level platformer, built in Python and Pygame.
I built a modular game with movement physics, buffered controls, collision detection and saved personal bests. Its 2,000+ lines of code included a level system, asset caching and save files. By caching and reusing loaded assets, I cut level-loading time from 650 ms to 170 ms.
I’m studying Maths & Computer Science at UCL. I like the bit where an abstract idea becomes something you can actually click, play, or learn with.
Outside these projects, I lead technology at UCL Hedge Fund Society. I’ve also taught 20+ beginners how to build and evaluate models. Making things is fun. Helping someone else make their first thing is pretty great too.
