What I'm doing now

I'm at ONDEWO working on AI4Emergency, an Austrian FFG-funded research project on AI-driven communication technology for emergency operations.

I finished my BSc in Artificial Intelligence at JKU Linz ahead of the standard duration, with a thesis on deep learning for language-based audio retrieval.

The learning that is still happening happens in public, in three repositories I keep pushing to:

• TempBench

  whether audio-language models can answer trivial questions about event order and timing

• Tython

  a statically typed toy language, its standard library and its toolchain

• nanoBeard

  a small language model trained from scratch, then supervised fine-tuned

I taught AI programmes at several coding schools as a freelancer — LLMs, multi-modal models and frontend development. That run ended in February 2026 when I moved to ONDEWO, and I am not currently teaching a cohort. Explaining things is still the part of the job I enjoy most, so this is a pause rather than a full stop.

The contact form is the reliable channel.

Last updated: 2026-08