Cyrus Nikolaidis

I build systems that help make AI safer — guardrails, evaluations, and monitoring for large language models.

I'm a Tech Lead at Meta's Superintelligence Labs, working on AI trust monitoring. Previously I led efforts to incubate the use of ML for privacy and security challenges at Meta.

Much of our work has been open-sourced and widely adopted by the community: CyberSecEval, LlamaFirewall, PromptGuard, and LlamaGuard.

Before all this, I worked on recommendation systems — I built the first version of Instagram's feed ranking and worked on Facebook search.