Niklas Stoehr
Research Scientist at Google DeepMind
Zurich, Zurich, Switzerland
Summary
👤
Senior🎓
Top SchoolNiklas Stoehr is a research scientist based in Zurich specializing in language model interpretability to drive efficiency and reliability gains both before and after training. With nine years of experience spanning industry research roles at Google DeepMind, Google, Bloomberg, IBM and academic stints at ETH Zürich and UCL, he blends rigorous ML/NLP theory with hands-on applied work. His PhD-level focus on applied and interpretable machine learning is complemented by internships and collaborations at leading labs, giving him deep exposure to production-scale model challenges. Notably, he pairs cross-disciplinary training—from industrial engineering to web science and a year at Tsinghua—with practical research that targets making large models more understandable and cost-effective.
9 years of coding experience
1 year of employment as a software developer
BSc, Industrial Engineering and Management, BSc, Industrial Engineering and Management at Technische Universität BerlinDoctor of Philosophy - PhD, Machine Learning and NLP, Doctor of Philosophy - PhD, Machine Learning and NLP at ETH ZürichOne-year Visitor (BSc), Computer Science, One-year Visitor (BSc), Computer Science at Tsinghua UniversityUniversity College London
German, English, Chinese, French
Stackoverflow
1reputation
0reached
0answers
0questions
Github Skills (21)
acl9
computational-linguistics9
phonetics9
python8
deep-learning8
gpu8
machine-learning8
numpy7
natural-language-processing7
gpu-acceleration6
neural-network6
tensorflow6
acceleration6
tensor5
autograd5
Programming languages (3)
Github contributions (5)
niklasstoehr/ordered-matrix-dirichlet
Jan 2023 - Dec 2024
We present an algorithm for constructing stochastic matrices with ordered latent states to circumvent label switching and improve interpretability when modeling international relations with dynamical systems and topic models.
Contributions:1 PR, 10 pushes in 1 year 11 months
niklasstoehr/ordinal-conflict-intensity
Oct 2022 - Nov 2024
Contributions:4 pushes, 1 branch in 2 years 1 month
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.