How scoring works for AI engineers
Being assessed by something you cannot see is a reasonable thing to dislike. So here is what the score is built from, what it is compared against, and which parts of it you can move.
This is the same information we give hiring companies, on their side of the site. Publishing one version for buyers and a thinner one for candidates would be the kind of thing you would rightly hold against us.
Ajai GeorgeLast updated
Five dimensions
- Depth
- Impact
- Leadership
- Innovation
- Trajectory
What the number is compared against
The five dimensions combine into a composite from 0 to 100, and you are placed in a percentile against engineers working in the same domain — drawn from the people we have already assessed.
Which is worth saying precisely, because it affects how much weight to give it: that percentile is your position within our assessed population, not a ranking against every AI engineer in the world. Nobody holds that dataset. Treat it as a read on where you sit among comparable engineers we have seen, and not as a global rank.
A low score is a starting position
The assessment names the dimension that is holding you back. If the gap is closable, that becomes a curriculum, and finishing it returns you to a fresh assessment rather than leaving the first one standing.
The legitimacy check works in your favour
A separate check runs first, on whether the profile is a real person's real work. Its verdict constrains the score, so a fabricated profile cannot present as strong.
If you are a genuine engineer with a quiet account, this is the part built for you. The check exists specifically to tell a thin-but-real profile apart from a manufactured one, rather than treating low volume as a red flag.
What we don't tell you
We will not publish how the five dimensions are weighted against each other, the formula that combines them, or the numbers a legitimacy verdict turns on.
You could fairly argue you are entitled to know, given it is your score. The reason we hold the line is that the moment those numbers are public, the rational move for everyone is to optimise against them instead of building — and then the score stops telling a hiring team anything, including about you. Withholding them is what keeps the thing worth having.