Saka Research

We are a machine learning research group led by İsmail İlkan Ceylan, with three homes: TU Wien and AITHYRA in Vienna, and the University of Oxford.

Our research is broad, spanning relational learning (graph learning, graph foundation models, and knowledge graphs), reasoning, generative AI, AI for mathematics, and AI for science. Across these areas, we combine theory and practice: we characterise what models can express, how they generalise, and how well they transfer across tasks and domains, and we use these insights to design new methods from first principles.

Many of our applications come from the sciences. How soluble is a molecule? Which genes are linked to a disease? How does a protein fold? Questions like these rest on structured, relational data, and together with our colleagues at AITHYRA we pay particular attention to the life sciences, with the aim of accelerating scientific discovery.

Saka /saˈka/, “sah-KAH”, is Turkish for the European goldfinch. Finches hold a special place in science: in The Voyage of the Beagle, Charles Darwin noted how the beaks of the Galápagos finches varied from island to island, an observation that became one of the best-known examples of evolution by natural selection.