Analysis toolkit for VEST, SNU's spherical tokamak — and the data platform around it. Diagnostics land in an OMAS-HSDS database, and from there the modelling runs itself: EFIT/CHEASE equilibria and DCON/RDCON stability, automated across the shot archive. Running it end to end turned up something I liked — VEST tends to be limited by current-driven instabilities rather than by pressure.

Energy-principle decomposition inside the General Perturbed Equilibrium Code — Fortran · Julia. A tokamak holds plasma on nested magnetic surfaces, and the field is supposed to stay on them. Measure a wobbling plasma the obvious way and the field looks like it leaks straight out through the surface — which would be alarming, if it were real. It isn't: you measured on the surface the plasma used to have. The effective field C = δB + (ξ·n̂)(μ₀j×n̂) is that same field read off the surface the plasma actually moved to, and the leak drops by nine orders of magnitude.

[Not fully implemented!]CFD-backed F1 front wing aerodynamics — parametric geometry, OpenFOAM RANS, and an interactive aero map. Turns out tokamaks aren't the only thing where the shape decides whether it works.