Sanjay Kairam
Measurement & Evaluation for AI and Sociotechnical Systems
Most of what matters in AI and sociotechnical systems resists easy measurement — that’s where I work. At Simile, that means Evals for frontier models that simulate and predict how people behave; previously at OpenAI, human evaluation methods for frontier models, and the same lens at Reddit and Twitch: belonging, moderation, community health. Stanford CS PhD, 1,700+ citations.
Currently
At Simile, working out what evaluation means when the model’s subject is a person — and how the choices we make there shape what these systems actually learn to do.
Selected Publications
How Conversational Structure and Style Shape Online Community Experiences
Galen Weld, Carl Pearson, Bradley Spahn, Tim Althoff, Amy X. Zhang, Sanjay Kairam
ICWSM 2026 · 2026
An analysis of how conversational structure and style relate to sense of community in online spaces, using large-scale behavioral and survey data from Reddit.
Post Guidance for Online Communities
Manoel Horta Ribeiro, Robert West, Ryan Lewis, Sanjay Kairam
CSCW 2025 · 2025
A novel approach to community moderation that proactively guides users' contributions using rules that trigger interventions as users draft posts.
How Founder Motivations, Goals, and Actions Influence Early Trajectories of Online Communities
Sanjay Kairam, Jeremy Foote
CHI 2024 · 2024
A mixed-methods study of how founder motivations, goals, and early actions shape the growth trajectories of online communities on Reddit.
Sanjay Kairam, Melissa C. Mercado, Steven A. Sumner
CSCW 2022 · 2022
A social-ecological model of sense of virtual community in Twitch livestreaming, examining individual, interpersonal, and community-level factors that predict belonging.