Mammogram quality assessment
MSc Computer Science · In progress
Automated PGMI scale classification using a ConvNeXt ensemble to assess mammogram image quality with deep learning.
My research interests are LLM training, evaluation, and reliable reasoning. My current public experiments study how language models reason and how to evaluate their behavior. Earlier, I worked on deep learning for medical image analysis at the University of Innsbruck.
Ledger-of-Thought
Independent experiment · July 2026
I evaluated Qwen2.5-7B-Instruct on a bookkeeping task with reasoning depths of 4, 8, 16, and 32, using 32 tasks per condition and depth. The experiment compares chain-of-thought prompting with an external state ledger and ablations of its checking mechanism.
At depth 32, the ledger condition reached 50.0% accuracy versus 28.1% for the chain-of-thought baseline. Removing the checking mechanism produced the same accuracy at every tested depth: the checks added no measured accuracy in Stage A.
These preliminary results cover a single model and task family. This is an inference-time evaluation using a fixed pretrained model. The report includes the experimental setup, confidence intervals, ablations, and limitations; code and recorded results are public.
University of Innsbruck
Kidney & tumor segmentation
BSc Computer Science · 2017–2020
Automatic semantic segmentation of kidneys and kidney tumors in medical images, combining YOLO-based localization with U-Net segmentation.
From research to practice
At Heliotherm, I applied machine learning to predictive maintenance for more than 4,000 heat pumps. At Accemic, I contributed to the EU-funded COEMS and TRISTAN research programs.