Vienna University of Economics and Business
BSc
Top 1% among 3000 colleagues while working part time.
Machine Learning Researcher
Credentials
Research Projects
01
ICLR 2026
CogniLoad
Separating intrinsic load, extraneous load, and length demand in LLM reasoning
Explore02
NeurIPS 2025 Efficient Reasoning Workshop
Decomposing Reasoning Efficiency
A workshop version of the efficiency accounting idea
Explore03
Preprint
Beyond Accuracy
An accounting protocol for why reasoning is inefficient
Explore04
PhD research
Structured Universal Latent Transport
Transporting hidden states across LLMs through an anchored universal latent space
ExploreCurrent work
I work at Integreat and at the Arctic University of Norway in Tromsø. My research centers on reasoning in LLMs: how to generate controlled tasks, how to account for token efficiency, and how to build representation transport between different models.
I currently research universal representations of hidden states among different LLMs, reinforcement-learning meta-policies for reasoning trajectories, and self-improving multi-agent communication protocols.
Service
I reviewed for
TMLR 2026NeurIPS 2026ICML 2026NLDL 2026NeurIPS 2025 Efficient Reasoning Workshop