Nikolaus Demmel
Machine Learning Engineer at Apple
Greater Munich Metropolitan Area Germany
Summary
🤩
Rockstar🎓
Top SchoolNikolaus Demmel is a Machine Learning Engineer based in the Greater Munich area with 16 years of software and research experience spanning industry and academia. He holds a strong research background in visual SLAM from a PhD at TUM and now applies that expertise at Apple, bridging cutting-edge algorithms with production-quality systems. Nikolaus is an active open-source contributor to renowned C++ and robotics projects (e.g., Pangolin, Ceres Solver, Sophus, Kalibr and rosdep), with a particular knack for build systems, cross-platform portability and numerical robustness. His contributions reflect a blend of algorithmic rigor and pragmatic engineering—improving calibration, optimization residuals, and build automation—rather than only high-level model tuning. Colleagues know him for meticulous code quality work (clang-format, dependency fixes) and for making research-grade vision tools more maintainable and widely usable.
16 years of coding experience
3 years of employment as a software developer
KTH Royal Institute of TechnologyMaster’s Degree, Computer Science, Master’s Degree, Computer Science at Technische Universität MünchenInternational Baccalaureate Diploma, International Baccalaureate Diploma at Sevenoaks School
Stackoverflow
379reputation
6kreached
1answer
6questions
Github Skills (45)
algorithm10
algorithms10
opencv10
c-language10
robotics10
geometry10
package-management10
python10
ceres-solver10
command-line-interface10
cmake10
clang-format10
plot10
homebrew10
camera10
Programming languages (18)
Github contributions (5)
tum-vision/LDSO
Sep 2018 - May 2020
DSO with SIM(3) pose graph optimization and loop closure
Role in this project:
Back-end Developer & Software Architect
Back-end Developer & Software ArchitectContributions:42 commits, 7 PRs, 16 pushes in 1 year 7 months
Contributions summary:Nikolaus contributed to the core functionalities of the LDSO project, fixing bugs related to argument parsing, memory allocation, and compiler warnings. They updated the code to work with different OpenCV versions and used `std::isnan` for robust error handling. The user also refactored the code to improve efficiency and maintainability, addressing issues and ensuring correct execution across various platforms.
structure-from-motionpose-graph-optimizationvisual-odometrysimoptimization
ros-infrastructure/rosdep
Feb 2014 - Nov 2018
rosdep multi-package manager system dependency tool
Role in this project:Back-end & DevOps Engineer
Back-end & DevOps EngineerContributions:22 commits, 4 PRs, 7 pushes in 4 years 9 months
Contributions summary:Nikolaus contributed significantly to the `rosdep` tool, adding new functionalities such as the `--skip-keys` and `--as-root` options, enhancing user control over dependency management. They also implemented Homebrew support, demonstrating expertise in package management. Furthermore, the user fixed permission issues and improved error messages, improving the tool's usability and robustness.
manager-systemmulti-packagepackage-manager
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.