Experiments

Things I build to find the edges of what AI agents can do, model evals, prediction, and live market bots. Some run here, some live on GitHub.

Arena

AI Model Comparison

Blind, side-by-side comparison across frontier models, generate, compare, and reveal which model actually won.

Open →Prediction

IPL Playoff Predictor

Surfaces playoff scenarios, decision impact, and net-run-rate tiebreakers as the IPL season unfolds.

Open →Markets

Polymarket Arb Bot

Within-market arbitrage on Polymarket binary markets, buys both sides when YES + NO dips below the threshold to lock in risk-free spread.

Open →Trading

Hyperliquid Trading Agent

A Claude-driven perps agent that reads technical indicators across 229+ markets and trades with hard-coded safety guards.

Open →