ai-elo-ranker

How it uses Jev

⚡ AI Elo & TrueSkill Ranker (Jev, Open-Weight Models & Bayesian Swiss Matchmaking)

Highlights

• $O(N \log N)$ Swiss-System Matchmaker vs $O(N^2)$ Round-Robin :
• Instead of comparing all 50,000,000 pairs in a 10,000-item set, the Swiss matchmaker pairs items with adjacent Elo ratings who haven't yet faced each other.
• Matches with equal expected win probabilities ( 50/50) yield maximum information gain ( 1 bit of entropy reduction per match).
• Reduces a 10,000-item tournament from 50 million duels down to 70,000 matches!
• Real-Time Visual Web Dashboard :
• Battle Arena : Dynamic dueling cards with Candidate A vs B, pulsing neon VS orb, winner slam animations, and floating +Elo / -Elo delta bubbles.

Getting started

1. Set Your Environment Ensure your .env contains your TypeSafe API credentials (optional if using local models or mock judge):