Arena is a community platform where millions of users benchmark frontier AI models against real-world tasks and vote on outputs—creating a public leaderboard grounded in production feedback rather than synthetic benchmarks. The stack is heavy on data infrastructure (Spark, Airflow, PostgreSQL, low-latency event streams) and security tooling (CrowdStrike, Cloudflare One, threat modeling), reflecting two acute scaling pressures: maintaining leaderboard integrity under adversarial conditions and building pipelines that ingest millions of evaluation signals at scale.
Arena, founded in 2025 by researchers from UC Berkeley, operates a community-powered platform for evaluating AI model performance in real-world settings. The platform aggregates feedback from millions of builders and researchers who test frontier models and vote on responses, feeding a public leaderboard. The company is headquartered in San Francisco and employs 51–200 people, with active hiring concentrated in engineering, data, and research roles at senior levels. Current product work centers on scalable evaluation infrastructure, platform security features, and real-time data pipelines to support the leaderboard's throughput and integrity.
Python, PyTorch, and Pandas for model work; Spark and Airflow for data pipelines; PostgreSQL and Supabase for persistence; Next.js and React for frontend; Cloudflare and CrowdStrike for security and infrastructure.
Scalable real-time data pipelines, low-latency event infrastructure, platform-wide security features, threat modeling, leaderboard integrity measures, and targeted sourcing for user feedback signals.
Other companies in the same industry, closest in size
Arena's technology stack, projects, and hiring signals are inferred from public hiring and company data — career pages, public listings, and company web presence — then clustered and de-duplicated. Figures are estimates that refresh over time. Read our full methodology →
This is not an official vendor or customer list. It is a technology-adoption signal inferred from public data, intended for B2B research.