About

I'm an Applied Scientist in Amazon's AGI team. I have a background in academic research and AI engineering and aim to translate cutting-edge AI research into useful applications.

In my academic research, I worked at the intersection of AI and biological science, mostly in London and Oxford (UK), but also in Germany. I investigated the nature of flexible, generalisable, and compositional intelligence in algorithms and biological data, focusing on reinforcement learning, Bayesian Inference and pattern analysis methods.

As an AI Engineer, I worked as a project lead and developer building Gen AI solutions in production. I also presented this work frequently in talks or blogs with a focus on Agentic AI (e.g. graph-based reasoning in agentic AI, Agentic Retrieval Augmented Generation (RAG), or the business value of Agentic AI) and reinforcement learning.

Experience

Feb 2026 – present: Amazon AGI | Sr. Applied Scientist, AGI | Berlin, Germany

Jan 2023 – Jan 2026: Alexander Thamm GmbH | Principal AI Engineer & Consultant | Munich, Germany

  • Role: Project lead & Gen/Agentic AI developer building production-grade ML solutions
  • Focus: Full-stack Agentic-AI applications, RAG chatbots for knowledge management, generative systems for automated reporting, forecasting and time-series analysis

Apr 2021 – Dec 2022: University of Tübingen | Postdoctoral Researcher, AI Center | Tübingen, Germany

  • Focus: Generalisable representation learning in artificial and biological intelligence (reinforcement learning, MCTS, graph neural networks)
  • Supervisor: Peter Dayan, FRS.

Apr 2017 – Mar 2020: University of Oxford | Postdoctoral Researcher | Oxford, UK

  • Focus: Compositional inference in reinforcement learning and computational vision, online and offline replay underwriting neural computations
  • Supervisor: Timothy Behrens, FRS.

2013 – 2017: University College London & University of Salzburg | PhD, Computational Neuroscience (Dist) | London, UK

  • Focus: Bayesian Decision Theory, Bayesian models of neural function, reinforcement learning
  • Supervisors: Karl Friston, FRS, Ray Dolan, FRS., Martin Kronbichler, Prof.

2012 – 2013: University College London | MRes, Cognitive Neuroscience (Dist.) | London, UK

  • Focus: Bayesian models of decison-making and neural function

2008 – 2017: University of Salzburg | BSc, Mathematics & BSc, Psychology (Dist.) | Salzburg, Austria

  • Focus: Bayesian parameter estimation, decision-theory