Essays on the systems I build — architecture, controls, and the discipline underneath.
2026-07-20
Research code rots because every new experiment quietly becomes a new copy of the code. The fix isn't discipline — it's making the copy structurally unnecessary. On rebuilding a lab platform around config-composition over orthogonal axes.