Tamás Vörös is a Senior Data Scientist at SophosAI, where he explores how machine learning and large language models can make cybersecurity smarter and safer. His recent projects include developing ways to harden AI models against jailbreak attacks (LLM Salting) and remove hidden backdoors (LLMBotomy).

Born and bred at SophosLabs straight out of university, Tamás has worked across web and spam protection, threat intelligence, and applied AI research. Over the years, his focus has shifted from hands-on detection systems to advancing the safety and interpretability of modern language models.

He regularly presents his research at conferences such as CAMLIS, Black Hat Europe, and Bsides. Tamás holds a Master’s degree in Computer Science from Eötvös Loránd University and is currently studying Psychology at Pázmány Péter University.