Adrian Hill
Berlin, Germany
Summary
🤩
Rockstar🎓
Top SchoolAdrian Hill is a Berlin-based machine learning PhD student with nine years of engineering experience bridging research, scientific ML, and maintainable code practices. He contributes to notable open-source projects—helping DiffEqFlux.jl generalize multiple-shooting solvers and improving documentation and tooling in jax-md and innvestigate—demonstrating both algorithmic insight and attention to code health. His background in mechanical engineering (M.Sc., TU Berlin, exchange at KAIST) informs a pragmatic approach to physics-informed ML and differentiable simulation. Colleagues rely on him for clean, well-tested solver integrations and readable documentation, a combination that helps move research prototypes toward robust, reproducible code.
10 years of coding experience
Master of Science (M.Sc.), Engineering Science, 1.2 (German GPA), Master of Science (M.Sc.), Engineering Science, 1.2 (German GPA) at Technische Universität BerlinExchange Semester (M.Sc.), Mechanical Engineering, Exchange Semester (M.Sc.), Mechanical Engineering at 한국과학기술원(KAIST)
German, French, English, Korean
Github Skills (22)
ode10
python10
ordinary-differential-equations10
code-standards10
machine-learning10
standardized10
standardization10
ml10
mle10
differential-equations10
neural10
coding-style10
scientific-machine-learning10
documentation10
julia10
Programming languages (14)
Github contributions (5)
albermax/innvestigate
May 2021 - Jan 2023
A toolbox to iNNvestigate neural networks' predictions!
Role in this project:
Back-end Developer
Back-end DeveloperContributions:6 releases, 1 review, 198 commits in 1 year 8 months
Contributions summary:Adrian primarily focused on code formatting and standardization within the repository. Their commits involved applying black and isort to the codebase, which indicates a focus on code style consistency and import organization. These changes were applied across various files, including core analyzer components and example code, suggesting a role in maintaining the codebase's readability and maintainability. The user also updated the documentation by fixing a typo and removing a module.
neural-network
SciML/DiffEqFlux.jl
Apr 2021 - Jun 2021
Pre-built implicit layer architectures with O(1) backprop, GPUs, and stiff+non-stiff DE solvers, demonstrating scientific machine learning (SciML) and physics-informed machine learning methods
Role in this project:ML Engineer
ML EngineerContributions:11 commits, 7 PRs, 13 comments in 2 months
Contributions summary:Adrian's contributions primarily focused on enhancing the `multiple_shoot` function within the `diffeqflux.jl` repository. Their work involved improving the solver's flexibility by making it agnostic to specific solvers and generalizing the continuity loss calculations. They also added features, such as a `sensealg` kwarg and general solver kwargs, and they improved the testing and documentation of the multiple shooting functionality.
machine-learningscientific-machine-learningneural-odeneural-sdeneural-pde
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.