Armin Thomas

Armin Thomas is the co-founder of Radical Numerics. They previously served as a Senior ML Scientist/Engineer at Liquid AI (2024-2025), where they developed STAR—Synthesis of Tailored Architectures for Hyena Edge, a convolution-based multi-hybrid LLM for edge devices with best-in-class latencies and memory usage. They conducted research at Stanford University as the Ram and Vijay Shriram Data Science Fellow (2021-2024), focusing on machine learning methods for training AI systems on high-dimensional neuroimaging data, and they were a Visiting Researcher at Caltech in 2014. They earned a Doctor of Science in Artificial Intelligence and Machine Learning from Technische Universität Berlin (2015-2020) summa cum laude, a Master of Science in Cognitive Neuroscience from Freie Universität Berlin (2012-2014), and a Bachelor of Science in Psychology from Freie Universität Berlin (2009-2012).