Dr Aleksander Zieliński
Postdoctoral Fellow in the MCNC Lab. Developing neuro-symbolic algorithms that combine foundation model representation learning with automated mathematical theorem proving.
Dr Aleksander Zieliński
Senior Postdoc in Neuro-Symbolic Computing & Formal Reasoning
Department of Computer Science & Intelligent Systems · Laboratory for Machine Cognition & Neural Systems (MCNC Lab)
Academic Biography
Dr Aleksander Zieliński is a Postdoctoral Research Fellow in the Laboratory for Machine Cognition & Neural Systems at BBQ Institute. He completed his undergraduate studies in Mathematics and Computer Science at the University of Warsaw and earned his D.Phil. (Ph.D.) in Computer Science from the Department of Computer Science at the University of Oxford under the supervision of the Automated Verification Group.
His research focuses on neural theorem proving, neuro-symbolic architectures, and automated invariant generation. At BBQ Institute, Dr. Zieliński develops neural systems that can interface directly with formal proof assistants such as Lean 4 and Coq, providing rigorous mathematical guarantees for machine-generated solutions.
Key Responsibilities & Leadership
- Core researcher on the NCN OPUS Grant on Neural Verification Architectures.
- Co-organizer of the BBQ Weekly Departmental Colloquium.
- Instructor for the Summer School on Formal Methods & Scientific Computing.
- Mentorship of MSc students in deep learning and mathematical logic.
Research Interests
- Neuro-Symbolic Integration & Neural SMT Solvers
- Automated Mathematical Theorem Proving & Lean 4 Architectures
- Efficient Attention Mechanisms & State Space Models (SSMs)
- Formal Language Theory & Differentiable Automata
Teaching & Courses
- Practical Neuro-Symbolic Programming with Lean 4 and PyTorch (CS-610)
- Data Structures and Advanced Algorithms (CS-201)
Selected Recent Publications
- Zieliński, A., Kowalczyk, J. (2026). Differentiable Proof Search in Large Mathematical Theory Spaces. ICML 2026.
- Zieliński, A., Nowak, K., Kowalczyk, J. (2025). Guided SMT Invariant Synthesis via Differentiable Logic Kernels. NeurIPS 2025.
- Zieliński, A. (2024). Exact Equivalence Checking of Quantized Neural Layers. ICLR 2024.
Office Hours & Student Advising
- Office hours: Fridays 10:00–12:00 (Room 3.09).
- Co-chair of the BBQ Institute Seminar Series 2026.