I design aerospace hardware
and the software behind it.

Senior research project, Oct 2025 to Mar 2026

An end-to-end design study of a small-scale hydrogen rotating detonation engine, carried from literature review to a working reacting-flow simulation. An RDE replaces the subsonic flame of a conventional combustor with a detonation wave travelling continuously around an annular channel.

I sized the combustor against the Bykovskii and Wolanski operability criteria, modelled it in Fusion 360, and built the reacting-flow case in CONVERGE CFD with detailed hydrogen chemistry, adaptive mesh refinement and periodic boundaries on an unwrapped annulus. The reference case reproduced a sustained rotating detonation wave, confirmed by front structure, periodic wrap-around without decay, and monitor-point pressure periodicity.

It is a demonstrated methodology and a qualitatively correct detonation, not a validated engine: wave speed was never compared against Chapman–Jouguet velocity, the cell-count cap ruled out a grid-convergence study, and the figures show that case rather than the custom geometry.

Combustor bore

75 mm

Annulus width

2 mm

Chemistry

SAGE, LLNL H₂

Solve time

~4 days, 4 cores

CONVERGE CFD, Fusion 360, Tecplot, ParaView, Python

View the repository

Independent project, Jul 2026 to Aug 2026

A physics-based RC aircraft design studio that outputs print-ready hardware. Give it an aircraft type, a mission, a cruise speed and a size box; it solves the coupled weight–aerodynamics–stability problem, returns five different aircraft, renders them in 3D, and exports CAD with servo pockets, wire runs, hinges, control horns and a magnet hatch already cut into the part.

Every number on screen is a real calculation: ISA atmosphere, NeuralFoil polars at RC Reynolds numbers, a spanwise strip model for the neutral point, tail-volume stability for tailed layouts. The CAD is the hard part: one watertight solid for the STL, a STEP assembly of named bodies, print-in-place captive-pin hinges, and every boolean existence-checked because OpenCASCADE will report success on an operation that did nothing.

Three generated airframes were run as full 3-D RANS cases on SimScale to check the internal physics against a solver that knows nothing about it; the small figures show the flying wing's streamlines and surface pressure at 20 m/s. One generated wing was printed in LW-PLA and test-flown.

Test suite

348 passed (v3.1 gate)

Aircraft families

6

Validation

3D RANS, k-ω SST

Best L/D in CFD

7.27, single AoA

Python 3.12, FastAPI, CadQuery / OpenCASCADE, NeuralFoil, three.js, SimScale

View the repository

Head-mounted thermal vision from a Raspberry Pi 4, a TOPDON TC001 USB thermal camera, and a pair of analog FPV goggles. The TC001 is built to plug into a phone; I wanted it to drive a headset and behave like an appliance — apply power, put it on, see heat.

The Pi pulls the 256×192 LWIR stream over V4L2 at 25 fps, renders it for seeing rather than measuring with local contrast enhancement and a light unsharp mask, and pushes NTSC composite out of the A/V jack. It boots straight into the viewer with no desktop, and the goggles' own input switch doubles as a mode switch between the drone feed and thermal. The clip is the goggles' own output; the two photos are the finished headset. Latency and battery life have not been measured yet.

Raspberry Pi 4, Python, OpenCV, V4L2, Shell

Repository

Independent project, Nov 2024 to May 2025

A CoreXY 3-D printer built from Rolohaun Design's open-source Rook 2020 MK2 and documented so the build can be reproduced or continued from another machine: the Klipper config, the CAD assembly, and an attributed parts list.

120 × 120 × 92 mm on a 2020 extrusion frame, BTT SKR Pico driving four TMC2209s over shared UART, Klipper and Mainsail on a Raspberry Pi, sensorless StallGuard homing on X and Y, and an HDX-Lite direct drive feeding a Bambu X1 hotend. Extruder calibration and resonance tuning are still open, and the repo says so.

Klipper, Raspberry Pi, BTT SKR Pico, Fusion 360

Repository