Christoph Reich
Ph.D. Student at Zuse School ELIZA
Darmstadt, Hesse, Germany
Summary
👤
Senior🎓
Top SchoolChristoph Reich is an ELLIS Ph.D. student researching unsupervised scene understanding and self-supervised representation learning across TUM, TU Darmstadt and the University of Oxford under leading advisors in computer vision. He brings eight years of experience spanning academic research, teaching, and an industry research internship at NEC Labs focused on optimizing image and video codecs for deep vision models. His open-source contributions include implementing advanced loss functions and metrics in Kornia and integrating Swin Transformer V2 into the timm repository, reflecting hands-on expertise in deep learning engineering for vision. Christoph has a strong biomedical imaging background with multiple peer-reviewed publications and practical experience building differentiable codec surrogates and robustness-focused model components. Based in Darmstadt, he combines rigorous theoretical work with applied engineering, notably adding support for non–16-divisible images in a popular computer vision library. He is known for bridging low-level vision concepts with scalable self-supervised methods that transfer across domains.
8 years of coding experience
Technischen Universität DarmstadtDoctor's Degree, Computer Science, Doctor's Degree, Computer Science at Technical University of MunichCollege Entrance Qualification (Fachhochschulreife), Electrical Engineering, 1.8, College Entrance Qualification (Fachhochschulreife), Electrical Engineering, 1.8 at Vocational College Leverkusen-Opladen
English, German
Github Skills (19)
pytorch10
loss-functions10
python10
pre-trained-model10
image-processing10
image-classification10
model-driven10
machine-learning10
model-building10
deeplearning-ai10
deep-learning10
neural-network10
computer-vision10
modeling10
model-driven-development10
Programming languages (5)
Github contributions (5)
kornia/kornia
Apr 2022 - Jan 2023
🐍 Geometric Computer Vision Library for Spatial AI
Role in this project:
ML Engineer
ML EngineerContributions:64 reviews, 3 commits, 11 PRs in 8 months
Contributions summary:Christoph primarily contributed to the implementation and testing of new loss functions and metrics related to image processing and computer vision, as evidenced by the addition of Lovasz-Hinge/Softmax losses, Welsch, Cauchy, Geman-McClure, and Charbonnier losses, and the Average Endpoint Error (AEPE) metric. They refactored existing code, added type checks, and incorporated test cases, indicating a focus on enhancing the functionality and reliability of the library's deep learning components. Furthermore, the user added support for images not divisible by 16 within the JPEG codec.
pytorchdifferentiablepythonvisiondeep-learning
huggingface/pytorch-image-models
Feb 2022 - Jul 2022
The largest collection of PyTorch image encoders / backbones. Including train, eval, inference, export scripts, and pretrained weights -- ResNet, ResNeXT, EfficientNet, NFNet, Vision Transformer (ViT), MobileNetV4, MobileNet-V3 & V2, RegNet, DPN, CSPNet, Swin Transformer, MaxViT, CoAtNet, ConvNeXt, and more
Role in this project:ML Engineer
ML EngineerContributions:10 commits, 3 PRs, 5 comments in 5 months
Contributions summary:Christoph contributed to the Swin Transformer V2 model implementation within the repository, focusing on adapting and integrating code from an external source. They added the Swin Transformer V2 model and then added modifications, including parameter name changes, input resolution checks, and functionality for resetting the classifier. The user also made additions to support the models which are specific to image classification tasks, including model functions.
mobilenetaugmixefficientnet-trainingvisual-recognitionnfnets
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.