Technical Program Management jobs

TPMs at AI companies herd the most expensive projects in software: training runs with hard compute reservations, data pipelines spanning three teams, and launches gated on safety reviews. The job is dependency management where one slipped date can strand a million dollars of reserved GPUs.

The strongest candidates speak enough ML to know why the eval suite blocks the release, and enough infrastructure to ask about cluster utilization without flinching. Labs and scaled AI companies hire this heavily; startups fold it into engineering management until they can't. Compensation tracks senior engineering more closely than people assume.