Matthias Minderer
Member Of Technical Staff at Microsoft AI
Zurich, Zurich, Switzerland
Summary
👤
Senior🎓
Top SchoolMatthias Minderer is a research-focused engineer with 12 years of experience building multimodal large language models and advanced ML systems, currently a Member of Technical Staff on the Microsoft AI Zurich team. He spent several years at Google DeepMind rising from AI Resident to Staff Research Scientist and is also listed as a Research Scientist at Google Brain, reflecting deep expertise at top AI labs. His academic grounding is a PhD in Neuroscience from Harvard and earlier degrees from ETH Zurich and Cambridge, fueling a research-first approach to model design. Matthias contributes to influential open-source projects—such as core improvements to Google Research’s Scenic JAX computer vision library—showing practical impact on tooling and datasets beyond pure research. Colleagues know him for bridging neuroscience intuition with scalable ML engineering, often improving loss functions, dataset integrations, and developer ergonomics. Based in Zurich, he maintains a public project portfolio at mjlm.github.io that highlights both experimental work and production-ready contributions.
12 years of coding experience
5 years of employment as a software developer
Master’s Degree, Neuroscience, Master’s Degree, Neuroscience at Eidgenössische Technische Hochschule ZürichBachelor of Arts (B.A.), Natural Sciences, Bachelor of Arts (B.A.), Natural Sciences at University of CambridgeDoctor of Philosophy (Ph.D.), Neuroscience, Doctor of Philosophy (Ph.D.), Neuroscience at Harvard University
Github Skills (13)
model-building10
computer-vision10
deeplearning-ai10
deep-learning10
jax10
data-integration10
modeling10
model-driven10
model-driven-development10
datasets10
data-set10
python9
transformers8
Programming languages (6)
Github contributions (5)
google-research/scenic
Jul 2021 - Jan 2023
Scenic: A Jax Library for Computer Vision Research and Beyond
Role in this project:
ML Engineer
ML EngineerContributions:36 commits, 64 comments, 43 issues in 1 year 6 months
Contributions summary:Matthias made several contributions related to the core functionality of the Scenic library for computer vision, specifically within the context of a Jax-based project. Their work included modifications to loss functions (e.g., setting `antialias=True` in dice loss interpolation), and the addition of new datasets or features like the "ref_coco" label map. They also integrated features like `nn.enable_named_call()` and the addition of a `ConfigDictWithAccessRecord` for testing. These changes indicate a focus on model development, dataset integration, and improving the development environment within the Scenic project.
computer-visionjaxdeep-learningtransformersvision-transformer
HarveyLab/Acquisition2P_class
Oct 2014 - Apr 2018
For managing 2P imaging datasets from preprocessing to activity trace extraction
Contributions:234 commits, 86 pushes, 4 branches in 3 years 6 months
imagingpreprocessing
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.