Roki Seydi

Through line

AI systems encounter people through data — records, clinical notes, interaction logs. The assumptions embedded in those representations about who people are, how they behave, and what they need are almost never tested against humans as they actually are.

Research interest

I study computational human context: how lived experience, relational history, and social position can be encoded without flattening people into static attributes or generic personas, and how context-grounded agents can expose failures in AI systems and sociotechnical environments that benchmarks cannot reach. Specifically, I research:

• how to build adversarial testing for clinical multi-agent systems before agent-based cyberattacks become a thing in clinical settings
• how to construct social agents that are psychologically plausible, auditable and based on human behaviours

My work

Healthcare is my primary domain. My work identified cultural confabulation — a previously uncharacterised failure mode in frontier model reasoning for healthcare — and built open-source evaluation infrastructure on the UK AI Safety Institute's Inspect platform to measure it systematically across frontier models. I am also deep in R&D for a research project at the intersection of AI safety, healthcare, agent psychology and cybersecurity.

Background

I am the inaugural Spärck AI Scholar at University College London, where I am completing an MRes in AI-Enabled Healthcare. I hold an MPH from Imperial College London and a BA from King's College London. I have presented at the Technical AI Safety Conference at Oxford, moderated at the Skoll World Forum, and review for ICML and NeurIPS.

Otherwise

I went to art school for five years and played in orchestras for nearly a decade. I grew up by Lake Como, Italy.