Julian Tatsch
Mitbegründer at Ceyeborg GmbH
Munich, Bavaria, Germany
Summary
🤩
Rockstar🎓
Top SchoolJulian Tatsch is a Munich-based technologist and co-founder with 11 years of experience bridging applied machine learning, computer vision research, and product-focused engineering. At BMW he progressed from PhD candidate in computer vision to data scientist, translating academic rigor into production ML solutions, and now co-leads Ceyeborg to commercialize those capabilities. His open-source work ranges from adapting PSPNet in TensorFlow/Keras for broader input sizes to improving UX and MQTT integration for IoT camera projects, reflecting a mix of deep model expertise and pragmatic full-stack improvements. Trained at TUM in Robotics, Cognition & Intelligence with a business background in TUM-BWL, he combines technical depth with commercial awareness. Colleagues describe him as someone who moves seamlessly between low-level model adjustments and user-facing interface polish — a useful quality for startups turning research into usable products.
11 years of coding experience
Diplomkaufmann, TUM-BWL, Diplomkaufmann, TUM-BWL at Technical University of MunichMaster of Science (M.Sc.), Robotics, Cognition, Intelligence, Master of Science (M.Sc.), Robotics, Cognition, Intelligence at Technische Universität München
Stackoverflow
57reputation
8kreached
0answers
3questions
Github Skills (19)
javascript10
mqtt10
mqttnet10
keras10
deep-learning10
tensorflow10
front-end-development10
html10
computer-vision10
python9
ui-design9
segmentation9
image-segmentation9
machine-learning8
css8
Programming languages (18)
Github contributions (5)
EliasKotlyar/Xiaomi-Dafang-Hacks
Dec 2017 - Dec 2022
Role in this project:
Full-stack Developer
Full-stack DeveloperContributions:2 reviews, 306 commits, 395 PRs in 5 years 1 month
Contributions summary:Julian primarily enhanced the user interface of the project, adding features for live view and various control elements. Their contributions involved the creation of HTML elements, the integration of JavaScript for functionality, and the implementation of responsive design elements, particularly switches for numerous functions, improving the overall user experience. They also added and fixed the MQTT functionality of the project, including autodiscovery features. The user also made several improvements to the general code style in the project.
Vladkryvoruchko/PSPNet-Keras-tensorflow
Aug 2017 - Dec 2017
TensorFlow implementation of original paper : https://github.com/hszhao/PSPNet
Role in this project:ML Engineer
ML EngineerContributions:7 commits, 7 PRs, 43 comments in 4 months
Contributions summary:Julian implemented and refined a TensorFlow-based implementation of the PSPNet model. They added support for different network configurations, integrated alpha blending, enabled Keras saving and loading functionalities, and addressed file path issues. The user also modified the model's architecture, particularly the pooling layer parameters, to accommodate different input image sizes. These changes demonstrate a focus on adapting the model for broader use cases.
deep-learningpspnettensorflow
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.