Ralf Herbrich
Ralf Herbrich studied computer science in Berlin and completed his doctorate in 2000. At Microsoft Research, he coauthored TrueSkill with Tom Minka and Thore Graepel, estimating individual skill and uncertainty from game outcomes. He held leadership roles at Microsoft, Facebook, Amazon and Zalando. In 2022 he returned to academic research and teaching at HPI and Potsdam, studying AI and sustainability. He left HPI effective 30 September 2026.
- Now at
- University of Potsdam
- Works
- 159
- Citations
- 9,619
- h-index
- 38
Publication numbers from OpenAlex, as of 14 October 2026.
Timeline
-
1997–2000
Study computation and learning
Herbrich completed a computer science diploma at TU Berlin in 1997 and a doctorate in 2000. He studied theoretical statistics and AI, which informed later probabilistic learning methods.
-
2006
Estimate skill and uncertainty
Herbrich, Tom Minka and Thore Graepel presented TrueSkill. The Bayesian model estimates skill from game outcomes, handles draws and team results, and tracks uncertainty. The paper also reports experience with the system in a commercial online gaming service.
-
2000–2021
Lead industrial research and applied AI
He held Microsoft research and leadership roles, followed by Facebook ranking work, Amazon machine learning leadership and Zalando AI leadership.
-
2022–2026
AI and sustainability research
He returned to basic research and teaching in 2022. His interests include approximate computing and Bayesian decision making. He left HPI effective 30 September 2026.
Sources and credits
Photo: hpi.de; creator not stated. License: Not stated by source. Source: hpi.de. Cropped to a square and resized. Ask us to remove or replace this photo.