Andreas Köpf
Founder & CEO at Roboteur
Münster, North Rhine-Westphalia, Germany
Summary
🤩
RockstarAndreas Köpf is an experienced open-source developer and AI engineer with over a decade building GPU-accelerated deep learning tooling, distributed inference systems, and web/cloud back ends. He has hands-on history contributing to foundational projects like Torch/PyTorch (implementing CUDA kernels and activations such as RReLU) and co-organizing large community efforts including Open-Assistant and LAION experiments. As a former Head of AI and inference engineer at Aleph Alpha and long-time architect of robotics and ML products at Xamla, he bridges low-level GPU/ CUDA work with higher-level system design and deployment. He co-founded community initiatives like the GPU reading group and helped establish Münster’s Warpzone hackerspace, showing a persistent focus on knowledge-sharing and grassroots technical leadership. Andreas combines deep technical craftsmanship (CUDA kernels, API/backend design, DB migrations) with a track record of organizing collaborative research and engineering communities. He’s currently focused on cognitive architectures and agent systems at Open-Thought, blending theoretical problem-solving with production-ready engineering.
11 years of coding experience
7 years of employment as a software developer
Github Skills (24)
apim10
pytorch10
fastapi10
activation-function10
loss-functions10
python10
back-end-development10
api-design10
gpu-programming10
c1110
c1710
deep-learning10
api10
neural-network10
cuda10
Programming languages (13)
Github contributions (5)
LAION-AI/Open-Assistant
Dec 2022 - Mar 2023
OpenAssistant is a chat-based assistant that understands tasks, can interact with third-party systems, and retrieve information dynamically to do so.
Role in this project:
Back-end Developer
Back-end DeveloperContributions:75 releases, 684 reviews, 333 commits in 3 months
Contributions summary:Andreas's commits primarily focused on back-end development within the OpenAssistant project. Their contributions included the initial implementation of back-end application features, specifically the creation of API endpoints for labelers and prompts. The user also worked on database schema migrations using Alembic, which included creating and modifying tables for core services like API clients, users, and data storage for prompt responses, showcasing expertise in database design and API development. They also added docker files.
language-modelpythonpartyassistantchat
torch/cunn
Oct 2015 - May 2016
Role in this project:Back-end Developer
Back-end DeveloperContributions:19 commits, 7 PRs, 8 comments in 6 months
Contributions summary:Andreas implemented the RReLU (Rectified Linear Unit with Randomized Leaky Rectification) activation function for the cunn library, adding the necessary CUDA kernels for forward and backward passes, and also added a test case. They also added conversions for various other activation and loss functions to integrate with the THCUNN library. This included the Abs, ClassNLLCriterion, DistKLDivCriterion, HardTanh, and L1Cost modules, among others, demonstrating an understanding of neural network components and their implementation on GPUs. Furthermore, the user refactored code by moving existing files for better project structure.
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.