Dmitry Kobak

Dr. Dmitry Kobak is a Group Leader in the Department of Data Science at the Hertie Institute for AI in Brain Health, University of Tübingen. His research focuses on unsupervised and self-supervised learning methods for analyzing high-dimensional biological datasets, particularly through contrastive learning, manifold learning, and dimensionality reduction techniques for visualization. He also applies his expertise to statistical forensics, including analysis of Russian electoral falsifications and modeling of Covid-19 excess mortality.

• Education:
  • PhD in Bioengineering, Imperial College London (2013)
  • MSc in Physics, University of Freiburg (2007)
  • BSc in Computer Science, St Petersburg ITMO University (2004)

Kobak’s work bridges theoretical advances in representation learning with real-world biological data challenges. He develops scalable dimensionality reduction algorithms like UMAP and t-SNE variants, optimized for single-cell transcriptomics and neural imaging datasets.

His scientific contributions are recognized through membership in the European Laboratory for Learning and Intelligent Systems (ELLIS) and association with the International Max Planck Research School for Intelligent Systems (IMPRS-IS).

• Key Affiliations:
  • Group Leader, Representation Learning and Data Visualization Group (2021–present)
  • Postdoctoral Researcher, University of Tübingen (2017–2021)
  • Postdoctoral Researcher, Champalimaud Centre for the Unknown (2013–2016)

Kobak’s lab actively develops open-source Python libraries for dimensionality reduction and data visualization, with applications spanning neuroscience, genomics, and social science data analysis.