Principal leadership — talent, hiring, and succession

Expected question

"How do you hire and grow an AI platform org so the architecture survives if you leave?"

Variant forms

"What do you look for in Staff AI engineers vs Senior application engineers?"
"Tell me about growing someone into owning a control plane."
"How do you avoid a bus factor of one on the gateway / eval platform?"
"When do you hire specialists vs generalists for agent systems?"

Executive summary

30-second thesis

I'd hire for judgment under ambiguity and production scars — not for who can recite RAG diagrams. Succession means a named backup owner for every invariant plane, with drills, not a wiki page.

2-minute answer

Bar for platform hires: can they choose a default, name what they'd refuse, and debug a failed mission from traces? Portfolio polish without operable stories is a soft no.

Growth: give ownership of a real gate (eval suite, gateway policy pack, tenancy quota ledger) with me as review, not as silent author. Mentorship that ends in "I still merge everything" failed.

Succession: for each critical system, a primary and a deputy who has paged on it. If I'm the only one who can revoke a tool, that's an incident waiting for vacation.

What I'd refuse: headcount theater — hiring five "AI engineers" with no shared control plane for them to inherit.

Evidence note

Use only verified mentoring/leadership stories from the evidence ledger (e.g. intern conversions, team build-outs). Don't invent span-of-control numbers in the room.

Skeptical follow-ups

"How fast should a deputy be ready?" Before the next promo cycle for that system — not "someday."
"What if product teams refuse platform standards?" Hire embedded partners who carry the invariant, or escalate with metrics — don't win by memo.

Staff+/Principal signal rubric

Senior: mentors individuals.
Staff+: multiplies via standards and ownership transfer.
Principal: designs succession for control planes; hiring bar matches the invariant model.