What the Talent Intelligence Score measures
A single number is only useful if you know what went into it. The Talent Intelligence Score is built from five dimensions, each read from work an engineer has already made public. Here is what each one is looking for.
Ajai GeorgeLast updated
Five dimensions
- Depth
- Impact
- Leadership
- Innovation
- Trajectory
The composite, and what it is relative to
The five dimensions combine into a composite expressed from 0 to 100. On its own that number means little, so each engineer is also placed in a percentile against a peer group — engineers working in the same domain, drawn from those we have already assessed. Where a domain group is still small, the comparison widens to the full pool rather than pretending to a precision it does not have.
That is worth being exact about: the percentile is a position within the population we have scored, not a ranking of every AI engineer alive. Nobody has that dataset, and a vendor who implies they do is telling you something else.
Legitimacy comes first
Scoring a fabricated profile accurately would be worse than useless — it would launder it. So a separate legitimacy check runs against the same public footprint, and its verdict constrains the score: a profile that fails it cannot present as strong, however impressive the underlying repositories look.
The order matters. We establish that the work is real before asking how good it is.
What we don't publish
We will tell you what the five dimensions are and what each one reads for. We will not publish how they are weighted against each other, the formula that combines them, or the thresholds a verdict turns on.
That is not coyness. Knowing that we measure trajectory tells you what we value; knowing exactly how much it counts tells someone how to manufacture it. The first is a product explanation. The second is a recipe for gaming the score, and it would degrade the signal for everyone reading it.