Rita Sevastjanova

Rita Sevastjanova is a postdoctoral researcher at the Interactive Visualization and Intelligence Augmentation Lab at ETH Zurich. She is dedicated to advancing interactive visual methods for explaining Large Language Models (LLMs). Specifically, she focuses on making the complex inner workings of these models interpretable to human users. This includes exploring how contextual word embeddings represent linguistic knowledge and, critically, how they might encode social biases. In the context of the exhibition project for SNF Nostos 2026, Dr. Sevastjanova provides essential expertise in LLM-based visualization and interactive AI systems. She is instrumental in shaping both the underlying conversational interface and the creation of "visual fingerprints"—unique visualizations generated in response to visitor interactions—that help translate complex AI processes into tangible, accessible insights.