Computational Neuroscientist (M.Sc., Univ. of Freiburg) with a Computer Science background (B.Sc., LMU Munich).

I focus on bridging Active Inference and first-principles probabilistic inference with physical & thermodynamic computing architectures.

• Thermodynamic Reasoning — p-bit POMDP
  Simulating POMDP belief updating and energy-based planning across 16 coupled p-bits in JAX via double-well Langevin dynamics.
  Interactive 3D Live Demo

• Master's Research – Contextual Information Seeking for Active Inference
  Unifying perception, planning, and action in POMDPs on Forney-style factor graphs. This work derives closed-form message passing with custom Bethe and generalized free energy functionals to actively resolve contextual uncertainty through curiosity.

• Theory: Active Inference, Free Energy Principle, Factor Graphs (FFG), Message Passing, Langevin Dynamics
• Computation: Python, NumPy, JAX