Background & CV
Project
WYVERN
A helicopter whose arms are its landing gear
Hover a side · click to open
Login
Esc to close
I'm a mechanical engineering student who likes problems where the maths has to survive contact with a real machine. A year on a certified rotorcraft programme taught me how much of engineering is evidence rather than ideas, and that's shaped everything I've built since.
Outside the degree I found and lead a UAS competition team, design a rotorcraft to the same rulebook it competes under, and write software to test whether things I believe are actually true.
Graduate and placement work in rotorcraft and unmanned systems, medical device mechanisms, or quantitative research — the three fields this site is organised around.
I'm most useful where a design has to be justified rather than just drawn: loads, dynamics, sizing, and the testing that proves it.
Rotorcraft & UAS
Project
A helicopter whose arms are its landing gear
Competition
IMechE UAS Challenge, 2027 fly-off
Rotorcraft design, unmanned systems, and a year of loads and dynamics work on a helicopter in service.
A single main rotor helicopter with two symmetric arms that double as landing gear. The arms are the mission: take off, pick up a 500 ml can, place it on a marker and land, flown inside the IMechE UAS Challenge envelope. CAD massing model in SolidWorks, specification at revision 0.3, and a sizing calculator spanning rotor aerodynamics, mass and CG, power and arm loads.
A sub-10 kg autonomous aircraft for the 2027 fly-off, and the organisation around it: Leonardo Helicopters as sponsor, university backing, and a committee running sub-teams through the society's first season.
Rotor dynamics, structural dynamics and airframe loads within Aircraft System Integration.
Devices & prosthetics
Status
First project being written up
Mechanism design for the human body — the same actuation, compliance and loads problems as a rotorcraft arm, at a scale where comfort matters as much as strength.
This section is deliberately empty rather than padded. Send me the biomedical work you want shown and it will read like the aerospace page: a cell in the comb, a specification, and an honest account of what was tested.
Analysis & algorithms
Signal research
Six timeframes, full cost model, paper traded
Machine learning
Evolution strategies over thousands of stock-days
Quantitative signal research in Python, measured honestly — including the runs that lost money and what they proved.
Ranks a UK universe by a correlation-weighted momentum score across six timeframes, sizes positions equally and exits on a fixed take-profit or stop. Two evaluated runs put the signal at roughly break-even before costs and negative after them, with stamp duty alone accounting for about three quarters of the loss — which turned the project from signal tuning to cost structure.
Treats each stock-day as an episode with a fixed opening balance and trains a network by evolution strategies over thousands of cached episodes. Fitness rewards mean return and the fraction of green days while penalising volatility. A walk-forward split is the overfitting alarm: validation is compared against buy-and-hold, so failing to beat it counts as a finding rather than a bug.
More to come
A sixth field, waiting on a name.
Candidates: software and game development, product and mechanical design, or teaching. Whichever you want a third group of employers to see first.
Get in touch
Happy to talk about rotorcraft, unmanned systems, sponsorship for the team, or anything on this site.
Kana
Pick a script and a set, then answer as many as you can. Wrong cards come back round later in the same session.