PhD Researcher @ AutoML Lab of Uni Freiburg (Prof. Frank Hutter) | LMU Munich Alumni | Ex-Airbus

Hey everyone! I'm Tom and I work on methods for optimizing LLM-based systems at inference time. My research brings Automated Machine Learning (AutoML) to the configuration of multi-agent systems, enabling automated setup, evaluation, and optimization in settings with expensive evaluation.

• AutoML for Agentic Systems
  Multi-agent systems are difficult to build and scale with manual tuning. I develop methods to automatically configure, optimize, and evaluate them under realistic constraints.

• Prompt Optimization
  Prompt engineering can be brittle and hard to generalize. I focus on automated optimization approaches to improve performance while reducing inference cost and environmental impact.

• Agentic Data Science
  Building systems that move beyond assistance toward executing end-to-end data science workflows with minimal human intervention.

🌟 promptolution (EACL 2026, 120+ ⭐)
Scalable framework for automated prompt optimization using evolutionary search.

💻 Repository · 📄 Paper · 🎥 YouTube Summary

📄 CAPO: Cost-Aware Prompt Optimization (AutoML 2025)
Efficient prompt search via budget-aware evolutionary optimization (live version in promptolution!)

💻 Repository · 📄 Paper · 🎥 YouTube Summary

📄 CALIOPE: Calibration of Positional Encodings (EACL 2026)
Training-free positional encodings applied at inference time that can massively improve the utilization of long contexts.

💻 Repository · 📄 Paper · 🎥 YouTube Summary

Open to Research Internships (2027)
🔗 LinkedIn · Google Scholar · E-Mail