problem solving
Measuring instead of asserting
Instrumenting the thing and reading the numbers, rather than trusting that something is fast enough, correct enough, or true because it looks that way.
Applies to performance and to claims equally. A frame budget is settled with a profiler, a contrast ratio with the specification’s own formula, and a statistic by opening the source it is attributed to.
Evidence4
This site, as a career database
I pick accent colours against computed WCAG contrast rather than by eye, and every research document I write grades its own sources. That is what caught a set of widely repeated hiring statistics which do not appear in the primary source everyone attributes them to.
Microscopy image processing and 3D reconstruction
Filtering goes through FFT convolution rather than direct convolution. I instrumented the viewer's frame rate with a timer that writes a benchmark file, moving average in one column and instantaneous in the other, so the speed is measured rather than judged by eye.
A real-time renderer built from the maths up
Lighting is recomputed per pixel every frame, with quadratic attenuation and a spotlight cutoff evaluated for each light at each fragment. A shortcut anywhere in that loop shows up as a dropped frame, so I could not guess at whether it was fast enough.
Electronics and embedded tinkering
A circuit either works on the bench or does not, so the multimeter decides rather than the reasoning that led to it.