Sergii Kashubin

Sergii Kashubin is a Senior Research Engineer at Google DeepMind, where his current work focuses on protein function discovery and agentic science in collaboration with the European Bioinformatics Institute (EMBL-EBI). His interest in the intersection of biology and artificial intelligence began 15 years ago during his computer science studies in Ukraine. Sergii began his career as a software engineer in Google Search Quality before transitioning into applied research in natural language compositionality and computer vision. Driven by a long-standing passion for biology, he has spent the past four years working in the life sciences, specializing in AI for proteins – spanning from molecular dynamics to high-level function prediction. His research has been published in premier venues such as ICLR and Science Advances. He recentlty contributed to the release of 26,000 novel protein function annotations with interpretable evidence (ProtNLM via UniProt) and an open-source agentic skill suite for automating life science workflows.

Towards Efficient and Reliable Scientific Workflows for Breakthrough Research

Abstract: Tackling grand scientific challenges in an industrial research environment requires a unique convergence of long-term vision, massive compute infrastructure, and multidisciplinary collaboration. In this talk, we explore how scientific research is conducted at Google DeepMind and examine the emerging paradigm shift toward Agentic Science. We introduce the open-source Science Skills suite, an agentic framework that equips language models with quality-verified tools across genomics, proteomics, and scientific literature. Through real-world biological case studies, we demonstrate how structured skills allow efficient models to match or exceed the reliability of unassisted frontier models while achieving substantial token savings. Finally, we discuss open challenges in tool maintenance, benchmark design, and agent steerability.