Selected work where curiosity turned into measurement: find the structure, build the experiment, and make the result clear enough that someone else can continue it.
Semester thesis at Imperial and ETH: finite-parameter experiments for the Syzygy distinguisher, testing whether shortened dual alternant and Goppa codes separate from matched random binary codes.
Imperial, winter 2025: LIF spiking network with surrogate gradients decodes 2D cursor velocity from motor-cortex spikes vs a non-spiking baseline.
ETH TMU (Prof. Stephan), spring 2025: spectral DCM on pre-treatment fMRI to predict ketamine response (NIMH, n=26)—compact 4-node models outperformed larger DCMs for binary response (~73% accuracy).
ETH IfA (Prof. Dörfler), spring 2025: multi-agent informative path planning on incrementally revealed graphs—beam search with Voronoi-style coordination and a bitmap hidden-path mechanism.
ETH 3D Vision (2025): urban Gaussian splats → clean meshes (~10% Chamfer gain on TartanAir vs strongest baseline) for simulation-ready city geometry.
ETH TIQI bachelor thesis (Profs. Novotny & Home): in-lockbox network analyzer on Red Pitaya + multi-client stack, validated against PyRPL—closed-loop Bode without extra lab hardware.
ETH, spring 2024: FPGA-based control and measurement stack for superconducting-qubit experiments.