Prof. dr hab. inż. Janusz Kowalczyk
Director of BBQ Institute. Leading research in foundation model alignment, distributed neural verification, and neuro-symbolic reasoning architectures.
Prof. dr hab. inż. Janusz Kowalczyk
Director of BBQ Institute & Chair of Machine Intelligence
Department of Computer Science & Intelligent Systems · Laboratory for Machine Cognition & Neural Systems (MCNC Lab)
Academic Biography
Prof. Janusz Kowalczyk is the Director of the BBQ Institute of Advanced Science & Technology and Full Professor in the Department of Computer Science & Intelligent Systems. He received his M.Sc. and Ph.D. in Computer Science from the Warsaw University of Technology and completed postdoctoral training at ETH Zürich in the Department of Computer Science. In 2018, he received his D.Sc. (Habilitation) from the Institute of Fundamental Technological Research, Polish Academy of Sciences.
His research lies at the intersection of deep learning theory, neuro-symbolic integration, and verifiable distributed systems. Over the past fifteen years, Prof. Kowalczyk has authored over 90 peer-reviewed articles in top venues including IEEE TPAMI, JMLR, NeurIPS, ICML, and ACM Computing Surveys. He has served as Principal Investigator for multiple major grants funded by the National Science Centre (NCN), the Foundation for Polish Science (FNP), and the European Commission’s Horizon Europe framework.
Research Group & Laboratory
Prof. Kowalczyk directs the Laboratory for Machine Cognition & Neural Systems (MCNC Lab) at the Ochota Science Campus in Warsaw. The lab hosts 6 postdoctoral fellows, 12 doctoral researchers, and collaborates closely with institutions across Europe.
Current investigations focus on:
- Mathematical bounds and guarantees for large parameter foundation models.
- Formally verifiable inference protocols for safety-critical computational science.
- Hybrid architectures combining differentiable representations with symbolic theorem provers.
Key Responsibilities & Leadership
- Overall academic and strategic leadership of BBQ Institute of Advanced Science & Technology.
- Principal Investigator for NCN OPUS 24 Grant on Scalable Neural Verification Protocols.
- Supervision of doctoral dissertations in deep learning theory and verifiable computation.
- Co-chair of the Central European Academic AI Consortium (CE-AIC).
Research Interests
- Deep Learning Theory & Generalization Bounds
- Neuro-Symbolic Reasoning & Formal Proof Generation
- Distributed Computing & Verifiable Machine Intelligence
- Scientific Provenance in Large-Scale AI Systems
Teaching & Courses
- Advanced Neural Computing & Deep Architectures (CS-801, Doctoral School)
- Theory of Distributed Machine Learning (CS-504)
- Research Seminar in Machine Cognition (CS-900)
Selected Recent Publications
- Kowalczyk, J., Zieliński, A. (2026). Verifiable State Space Models for Deterministic Sequence Reasoning. IEEE TPAMI.
- Kowalczyk, J., Markiewicz, E., et al. (2025). Provenance-Aware Training Pipelines for Scientific Deep Learning. Nature Machine Intelligence.
- Kowalczyk, J. (2024). Scalable Verification for Distributed Foundation Models. Journal of Machine Learning Research (JMLR).
Office Hours & Student Advising
- Office hours: Tuesdays 14:00–16:00 (Room 3.18 or via appointment).
- Accepting 2 new PhD students for the 2026/2027 academic year.