How ranking works

No black box, no magical AI agent deciding who's talented. Every candidate gets a weighted score built from three inputs, always in the same order of importance.

1. Portfolio depth

Highest weight

Rewards documented process and iteration, not merely upload count. Projections are assessed on depth according to start, rough version, final, lessons learned segments. A single project broken into real stages outweighs ten unlabeled files. Staging can be done retroactively.

2. Puzzle performance

Second weight

Culture-fair, nonverbal reasoning puzzles: pattern completion, odd-one-out, sequence, and grid logic. Deterministically graded, no AI judgment call involved. This is a discovery mechanic, most valuable for students with thinner portfolios (newer, younger, or earlier in their journey). Skipping the arcade doesn't hide a student; it just means they lean fully on their portfolio.

3. Mentor endorsements

Lowest weight

Tied to real names and organizations; a supplementary trust signal. Deliberately the smallest input, since mentorship access itself tracks the same networks and proximity this platform is trying to route around.