Soroush Ebadian

PhD at the University of Toronto | AI Safety Fellow at Anthropic

Toronto, Ontario, Canada

About

Experience

  • AI Safety Research Fellow at Anthropic
    Jan 2026 - Present · 6 mos

  • Research Assistant at University of Toronto
    Sep 2020 - Jan 2026 · 5 yrs 5 mos

    Presented my work at top-tier conferences in AI and Economics and Computation such as NeurIPS, ACM EC, AAAI, IJCAI, AAMAS, and ACM EAAMO.

  • Research Fellow at Pivotal Research
    Jul 2025 - Aug 2025 · 2 mos

    Conducted research on AI safety through formal analysis of the multi-agent off-switch game, demonstrating that individual corrigibility may fail to compose in multi-agent settings.

  • Machine Learning Researcher Intern at RBC Borealis
    Sep 2024 - Jan 2025 · 5 mos

    Conducting research on deep multi-task learning models for time series data.

  • Visiting Researcher at Harvard University
    Sep 2023 - Dec 2023 · 4 mos

    Host: Prof. Ariel Procaccia