writing

Short positions on AI, data work and leading teams. Updated as I learn.

Modeled data, governed metrics, and team context: what AI needs from a data team, and how AI builds it.

When everyone can write code, the senior job is orchestrating AI. Model selection is the first call.

For high-volume judgment work, AI is better than people. When it isn’t, the gap is context.

AI should take the grunt work, not the job, and put analysts back at the table.

Leadership is mostly small moments, and a few lessons from running my own company.

A contamination-resistant eval of six frontier models on Medicare billing rules, closed book and open book.