Nils Lehmann
Visiting Researcher at University of California, Berkeley
Munich, Bavaria, Germany
Summary
🤩
Rockstar🎓
Top SchoolNils Lehmann is a PhD candidate and associate researcher at TUM with eight years of experience applying generative modeling and uncertainty quantification to large-scale spatio-temporal Earth observation data. He blends academic research with practical engineering—contributing to open-source projects like microsoft/torchgeo where he added dataset plotting, band selection, and multi-label integrations that streamline remote sensing workflows. Currently a visiting researcher at UC Berkeley collaborating on generative sea surface state estimation, he is adept at building complex data pipelines and reproducible models for geospatial applications. Based in Munich, he pairs rigorous data-science training with hands-on software development, and his background includes unexpected stints from collegiate athletics to working as a sous chef, reflecting adaptability and team-oriented problem solving.
8 years of coding experience
Bachelor of Arts - BA, Bachelor of Arts - BA at Rollins CollegeDoctor of Philosophy - PhD, Data Science for Earth Observation, Doctor of Philosophy - PhD, Data Science for Earth Observation at Technical University of MunichMaster of Science - MS, Master of Science - MS at University of AmsterdamHigh School, High School at Kieler Gelehrtenschule
English, German
Github Skills (11)
computer-vision10
pytorch10
pytorch-lightning10
python10
datasets10
matplotlib10
machine-learning9
data-visualisation9
data-visualization9
deep-learning9
data-visualizations9
Programming languages (4)
Github contributions (5)
microsoft/torchgeo
Dec 2021 - Jan 2023
TorchGeo: datasets, samplers, transforms, and pre-trained models for geospatial data
Role in this project:
Data Scientist
Data ScientistContributions:1 release, 342 reviews, 49 commits in 1 year 1 month
Contributions summary:Nils primarily contributed to the `microsoft/torchgeo` repository by adding plotting methods for multiple datasets, including the LoveDA, GID15, Cyclone, COWC, CV4A Kenya Crop Type, Levir, Zueri Crop, PatternNet, SEN12MS, Esri 2020, EuroSat, So2Sat, and CMS Global Mangrove Canopy datasets. These plotting methods visualize the data and are essential for inspecting dataset samples and model outputs. Moreover, the user implemented band selection, and integrated the multi-label classification dataset for the EuroSAT and So2Sat datasets, which is useful for the analysis of remote sensing data.
pytorchsamplersgeospatialdeep-learningearth-observation
nilsleh/torchgeo
Dec 2021 - Mar 2025
TorchGeo: datasets, transforms, and models for geospatial data
Contributions:785 pushes, 143 branches in 3 years 4 months
wpsgeospatialdata-managementdataminingdataset
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.