With a background in theoretical computer science, I work on formal methods and artificial intelligence. My research draws on logic, automata theory, and synthesis to study the verification, explainability, and interpretability of AI systems. I am part of the PEPR IA SAIF project and the FM4AI group at LMF.

Current research interests

• logical foundations of distributed and multi-agent systems

• formal methods for machine learning and agentic AI, including explainability and verification

• synthesis, planning, and grammatical inference

• ZipperGen, a Python DSL for developing reliable LLM-agent workflows (GitHub)

Recent publications and preprints

• Z. Xu, B. Bollig, M. Függer, T. Nowak, V. Le Dréau: Centralized Yet Scalable: Revisiting Policy Design for Cooperative Multi-Agent Learning, accepted at ECML PKDD 2026

• B. Bollig, M. Függer, T. Nowak: Provable Coordination for LLM Agents via Message Sequence Charts, accepted at ISoLA 2026

• B. Bollig: Deadlock-Free Parallel Regions for Projected Workflows, accepted at EXPRESS/SOS 2026

• B. Bollig: Causal Past Logic for Runtime Verification of Distributed LLM Agent Workflows, accepted at ICFEM 2026

• B. Bollig, M. Függer, T. Nowak, P. Zeinaty: Agent-Alternation-Free Epistemic Metric Temporal Logic with Past: Model Checking and Complexity, under submission

Recent activities

• talk on ZipperGen at the Annual PEPR IA Days 2026

• talk on explainability, verifiability, and the AI Act

• lecture notes on the theoretical foundations of neural network verification

• lecture notes on runtime verification from an epistemic perspective