Nicolai Wojke
Staff AI Software Engineer at Tensordyne
Berlin, Germany
Summary
🤩
Rockstar🎓
Top SchoolNicolai Wojke is a Staff AI Software Engineer in Berlin with 10 years of experience building production-ready perception and tracking systems for automotive and robotics applications. He holds a doctorate in Computational Visualistics and has progressed from academic research and PhD work on multi-object tracking to leading perception teams and shipping deployed models at companies like Vay Technology and Tensordyne. Nicolai's open-source contributions include substantial enhancements to the widely used deep_sort tracker—improving model architecture, inference freezing, and compatibility across TensorFlow/Python versions—which reflect his focus on robust, deployable ML systems. He combines hands-on model engineering (batch norm, dropout, residual blocks) with systems-level attention to detection/tracking data integrity, making him equally comfortable prototyping research ideas and hardening them for real-world use. Based in Berlin, he brings a measured blend of research rigor and product-facing pragmatism, often surfacing non-obvious trade-offs between accuracy and inference efficiency in deployed perception pipelines.
10 years of coding experience
13 years of employment as a software developer
Dr. rer. nat., Dr. rer. nat. at University of Koblenz and LandauArtificial Intelligence, Artificial Intelligence at University of Georgia - Franklin College of Arts and Sciences
German, English
Github Skills (9)
mask-rcnn10
faster-rcnn10
computer-vision10
machine-learning10
deep-learning10
tensorflow10
python10
model-optimization9
object-tracking9
Programming languages (2)
Github contributions (5)
nwojke/deep_sort
Feb 2017 - Jan 2019
Simple Online Realtime Tracking with a Deep Association Metric
Role in this project:
ML Engineer
ML EngineerContributions:29 commits, 19 PRs, 31 pushes in 1 year 11 months
Contributions summary:Nicolai primarily focused on modifying and enhancing the core deep learning components within the repository. Their contributions included significant updates to the model architecture, specifically involving batch normalization, dropout, and residual blocks. They also addressed TensorFlow version compatibility and implemented the freezing of the model for deployment, demonstrating a focus on optimizing the model for inference and real-world application. Furthermore, they adjusted detection and tracking components to ensure data integrity, and adapted the code for Python 2 compatibility.
realtimeassociationrealtime-trackingtrackingdeep-association-metric
nwojke/pymotutils
Aug 2017 - Aug 2018
Contributions:21 commits, 4 PRs, 15 pushes in 11 months
pythonpython-utilityobject-detectiontrackingmultiple-object-tracking
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.