Jonathan Klimesch
Doctoral Student Differentiable Programming & Detector Discovery at University of Tübingen
Munich, Bavaria, Germany
Summary
🤩
Rockstar🎓
Top SchoolJonathan Klimesch is a doctoral student and differentiable programming engineer based in Munich with eight years of experience bridging physics, machine learning, and software engineering. He designs AI-driven experiments and JAX-based simulators for detector optimization, bringing research-grade code into practical use at the Max Planck Institutes and now the University of Tübingen. His open-source contributions include substantive work on tum-pbs/PhiFlow, adding FLIP fluid dynamics features and visualization improvements that demonstrate applying differentiable PDE solvers to real-world simulations. As a founder of a learning-content search/recommendation startup, he combines product instincts with rigorous technical depth from dual degrees in physics and computer science. Colleagues value him for expanding action choices in complex systems — a philosophy reflected in his work to make simulation and detector design more controllable and searchable.
8 years of coding experience
3 years of employment as a software developer
Bachelor of Science - BS, Computer Science, Bachelor of Science - BS, Computer Science at Technical University MunichMaster of Science - MS, Physics, Master of Science - MS, Physics at Technical University of MunichHong Kong University of Science and Technology (HKUST)
English, German, Spanish
Github Skills (6)
pytorch10
machine-learning10
python10
deep-learning9
numpy8
tensorflow7
Programming languages (2)
Github contributions (5)
tum-pbs/PhiFlow
Nov 2020 - Apr 2021
A differentiable PDE solving framework for machine learning
Role in this project:
ML Engineer
ML EngineerContributions:11 commits, 4 PRs, 7 pushes in 5 months
Contributions summary:Jonathan made significant contributions to the `phiflow` project, a differentiable PDE solving framework for machine learning. Their work includes implementing features such as random number generation and comparison methods for geometric classes, along with refactoring and improving the extrapolation method. Further, they were involved in adding methods for implementing a FLIP simulation for liquids, including the implementation of viscosity options and the map_velocity_to_particles methods which showcases the application of the framework in fluid dynamics simulations. The user's work extends to improvements in PointCloud visualization and adding tests for the FLIP methods.
differentiablesolvingautomatic-differentiationdeep-learningpde-solver
StructuralNeurobiologyLab/MorphX
Oct 2019 - Aug 2020
Contributions:213 commits, 5 pushes, 1 branch in 10 months
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.