Hi, I’m Tom.
I'm a PhD researcher in machine learning at the University of Freiburg and the ELLIS Institute Tübingen. This is my corner of the web: a short CV and an occasional blog.
About
I work on prompt optimization and multi-agent system optimization in Frank Hutter's group — working toward the automated design and optimization of multi-agent systems. Before the PhD I was a data scientist at Airbus and studied statistics and data science at LMU Munich. I build in the open — my code, papers, and benchmarks are public.
I have opinions about what the future of agentic systems should look like — and mathematical and statistical evidence to back them up. I write about both on the blog.
The best way to reach me is by email: [email protected] — or find me on GitHub, Google Scholar, LinkedIn, YouTube.
Publications
- CANTANTE: Optimizing Agentic Systems via Contrastive Credit Attribution
NeurIPS 2026 (under review) · code - MO-CAPO: Multi-Objective Cost-Aware Prompt Optimization
arXiv preprint (2026) - Can Calibration of Positional Encodings Enhance Long Context Utilization?
Findings of EACL 2026 · code - promptolution: A Unified, Modular Framework for Prompt Optimization
EACL 2026 (System Demonstrations) · code - CAPO: Cost-Aware Prompt Optimization
AutoML 2025 · code
Full citations with author lists on the CV.
Software
promptolution — a unified, modular open-source framework for prompt optimization. Presented as a system demonstration at EACL 2026 — or watch the demo video.