AI quality / 2026-06-14
Enterprise AI needs a proof language
AI delivery is not only about building the feature. In enterprise systems, the hard work is often agreeing what proof different stakeholders will trust.
Read local versionWriting
Selected posts, kept readable outside the feed.
These are local, public-safe versions of LinkedIn writing pulled from my vault. They are ordered by portfolio relevance rather than feed chronology: AI proof, leadership paradoxes, AI-assisted judgment, and accountable engineering communication.
Engineering leadership / 2026-06-08
Leadership problems are often paradoxes
Good leadership is often not choosing one side of a tension. It is making both sides explicit enough that people can operate.
Read local versionAI-assisted judgment / 2026-05-24
AI is more useful as a room of perspectives
The useful AI move is not asking for one polished answer. It is choosing the lenses, constraints, and failure modes that should inspect the problem first.
Read local versionAccountability / 2026-05-11
Making engineering concerns useful
Software quality improves when objections are named clearly enough to inspect: violated property, evidence, risk, and decision needed.
Read local version