To AI Engineering Manager in 2027. Earn 2x or Die.

Two different jobs already share this title. One leads a team of human AI engineers building models, and companies are hiring for it right now. The other has no job postings yet: three to five mostly agentic teams, your span in humans stays around eight, and a fleet arrives on top of it.

We work through which one your org is already drifting toward, and what changes underneath you once coordinating people stops being the bottleneck and review capacity becomes it. Somebody has to decide what gets deep review, what gets sampled, and what gets trusted on reputation. In most orgs that call is made by default rather than by design.

The shape under pressure is the one most people hold today: one manager over one human team of six to nine, doing the job the way it was done in 2022. Two paths stay funded. Wider, across teams. Or back to the tools as a staff engineer directing a fleet.

You leave with the fork diagnostic: span, what you optimise, what breaks first, who funds it in 2027. Four rows, three columns, and you read the row you spend your week in, not the one on your business card. Plus the question to put to your director before someone answers it for you.

Built on 3,611 mentoring sessions since 2019. This talk is new and has not been delivered on any stage.