Alexandros Tzanakakis

Alexandros Tzanakakis is a first-year PhD student in Electrical and Computer Engineering at The University of Texas at Austin, where his research sits at the intersection of machine learning, computational genomics, and AI biosafety. He holds an integrated Master's degree in Electrical and Computer Engineering from the National Technical University of Athens, where he developed strong foundations in mathematics, machine learning, and large-scale computational systems. His current research focuses on the security and safety of genomic foundation models — investigating how dangerous biological capabilities encoded during pretraining can be suppressed, localized, and durably removed from open-weight models. His work includes the first empirical demonstration of capability locking and adversarial bypass in a genomic language model, showing that post-hoc safety interventions can be circumvented and motivating the development of more robust capability removal methods. He also contributes to research on training data poisoning in genomic models and agentic biosecurity risks in biological AI systems. Alongside this biosafety work, Mr. Tzanakakis develops long-context self-supervised genomic foundation models, with a focus on progressive scaling, training stability, and efficient architectures deployable on standard research hardware. His broader aim is to build genomic AI systems that are both capable and safe to release, contributing technical tools and empirical evidence to the governance of open-weight biological foundation models.

Personal Website

-

LinkedIn

LinkedIn

Twitter / X

-

LessWrong

-

EA Forum

-

Projects

No grants recorded.