Felix Köster
Forscher (Postdoktorand) at Saitama University (埼玉大学)
Saitama Prefecture, Japan
Summary
👤
Senior🎓
Top SchoolFelix Köster is a machine learning researcher and engineer with a PhD in theoretical physics, specializing in sequence models and time-series forecasting at the intersection of dynamical systems and deep learning. At Saitama University he designs and benchmarks novel neural architectures—such as reservoir–attention hybrids and causal-convolution transformer front-ends—that have produced multi-fold accuracy gains on chaotic and PDE-based forecasting tasks and consistent single-digit improvements on language modeling. He combines hands-on systems work (PyTorch, CUDA, C++, mixed-precision training and reproducible experiment pipelines) with theoretical analysis using dynamical-systems diagnostics to probe memory, stability, and model behavior. His projects span training GPT-style models at scale (7M–120M parameters on ~18B tokens) and shipping open-source tools like CTM-CIFAR Bench, LAERC, CCT and a C++ NEAT implementation. Fluent in turning physical intuition into practical ML solutions, he is seeking applied AI research or engineering roles focused on time-series and quantitative modeling.
9 years of coding experience
1 year of employment as a software developer
PhD, Physics, PhD, Physics at Technische Universität Berlin
English, German, Japanese
Github Skills (4)
regression5
delay4
machine-learning3
embedded2
Programming languages (1)
Github contributions (5)
Rincewind1989/MMF
Aug 2021 - Feb 2022
Contributions:26 commits, 28 pushes, 1 branch in 5 months
Rincewind1989/World-Simulation
Nov 2017 - Dec 2018
Contributions:22 pushes, 1 branch in 1 year 1 month
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.