PATRICIA PASKOV

Hello! I am a researcher at RAND and a Research Affiliate at Oxford Martin AI Governance Initiative. My work on evaluations and policy has been published at top machine learning conferences including ICML and FAccT as well as in policy venues including the World Bank, RAND, and Carnegie. I am also the Resilience section lead for the 2026 International AI Safety Report, a Working Group Chair at the EvalEval Coalition, and Faculty at the International Programme on AI Evaluation.

Previously, I built machine learning models at Condé Nast and designed and managed large-scale randomized controlled trials with The World Bank and Innovations for Poverty Action on topics including financial inclusion, mental health, education, and local governance.

I am a EU-US dual citizen with experience working and living across five continents in countries including Italy, Senegal, Thailand, Peru, and Paraguay. I studied Economics at Barcelona School of Economics and European University Institute, and Applied Economics at the University of Wisconsin-Madison.

I'm based in NYC and co-founded NYAIGS. I am keen to grow the city's frontier AI governance and safety community. If this interests you, let's chat.

Recent publications

Recommendations and Reporting Checklist for Rigorous & Transparent Human Baselines in Model Evaluations

Wei, K.L., Paskov, P., Dev, S., Byun, M.J. et al. | ICML (Spotlight), 2025

In Which Areas of Technical AI Safety Could Geopolitical Rivals Cooperate?

Bucknall, B., Siddiqui, S. et al. | FAccT, 2025

Methodological Challenges in Agentic Evaluations of AI Systems

Wei, K. et al. | ICML Technical AI Governance Workshop, 2025

Say hello!