ALL SPEAKERS
AT A GLANCE
Role
Software Developer
Organisation
cronn GmbH
Category
Community
ABOUT THE SPEAKER
I am a Software Developer and Machine Learning Engineer at cronn, with a focus on building local-first AI pipelines. I go deep into training and optimizing small open models so they run reliably on a single GPU instead of a cloud API, and I am happiest when a model gets smaller and better at the same time.
TALK
Achieving Reliable Automation: Quantifying Confidence with LLM Information Fusion
Reliable automation with Large Language Models (LLMs) requires not only accurate outputs but also trustworthy confidence estimates. Our work addresses the challenge of quantifying result reliability when extracting structured information from noisy, real-world inputs such as receipt images. We propose an information fusion approach that combines multiple LLM outputs and models their agreement as an assignment problem. By correlating items across independently generated results using similarity-based cost functions, we derive fused outputs and confidence scores grounded in cross-model consistency. The method filters hallucinations, resolves ambiguities, and produces a single high-quality result, enabling safer and more robust automation in high-stakes applications.
MORE VOICES
Other speakers at code.talks
See them live — and 100+ more.
One ticket, every track. Nov 4–5, 2026 at Kinopolis Hamburg.
Get your ticket





