Arena builds a platform to test how AI models perform on real work. Its product appears to combine public head-to-head comparisons, model rankings, and enterprise evaluation tools. The hiring mix points to a company moving beyond a single leaderboard into a broader evaluation stack: platform, data infrastructure, ML infrastructure, applied AI, research, product, design, security, and product solutions engineering. The customer set is likely both AI labs and companies choosing or deploying frontier models. The job copy says tens of millions of people use Arena each month to compare models and submit preferences, which Arena then turns into evaluation signals. In plain terms, Arena is trying to become the measurement layer for AI quality, reliability, and trust.