Hiring & growing AI capability
Build the bench, don't hoard the skill.
An AI Tech Lead grows the org's capability, not just their own. That means knowing what to hire vs upskill, what 'good' looks like in AI engineers, and how to build a durable bench so the program outlives you.
Key ideas
- 1
Upskill first: most AI capability is grown from strong existing engineers, not hired in. Hire to fill specific gaps (e.g. ML/eval depth, data engineering).
- 2
Screen for the right things: problem framing, evals/measurement instinct, debugging non-determinism, and judgment about when NOT to use AI — over trivia.
- 3
Create growth paths: champion → embedded AI engineer → mentor. Make the role attractive and recognized.
- 4
Avoid the hero trap and the single-point-of-failure: document, pair, and spread knowledge so capability is resilient.
- 5
Plan succession from the start: a second hub person and a champion bench mean the program survives turnover.
Buy vs build (people edition)
- Build: upskill curious, strong engineers via the guild, projects and pairing.
- Buy: hire for specific depth you lack (evals/ML, data, security) — usually a few key roles.
- Beware hiring 'prompt experts' with no engineering depth.
What 'good' looks like
- Frames problems and measures them (evals) before building.
- Comfortable with ambiguity and non-determinism; ships and iterates.
- Knows the limits — picks the simplest tool, including 'not AI'.
Watch
Do the work
0/4 · 0%Test yourself
What's the default way to build AI capability in an org?
27 chapters · progress saves automatically