I am currently a Senior Researcher (AI) at Software Competence Center Hagenberg (SCCH). My background is in mathematics, physics, and statistics, and my doctoral work, carried out within the S3AI project, focused on invariant geometric structures, notably convexity interpreted as space folding. I work across machine learning, neural-network geometry, mechanistic interpretability, and surrogate models for physical simulation, with additional experience in computer vision, natural language processing, and time-series modeling.

Research Landscape

How my research themes, papers, and projects connect. Hover or focus a keyword to preview it; select it to open its page.Tap a keyword to preview it, then follow its links.

Selected Publications

Journal and Conference Papers

TMLR 2026

LOCO: Local Orbit Consistency for Symmetry-Robust Neural Operators

Michał Lewandowski, Bernhard Heinzl, Roman J. Rainer, Bernhard Nessler, and Bernhard A. Moser

In Transactions on Machine Learning Research, 2026

AAAI 2026

Exploiting Space Folding by Neural Networks

Michał Lewandowski, Raphael Pisoni, Bernhard Heinzl, and Bernhard A. Moser

In Proceedings of the AAAI Conference on Artificial Intelligence, 2026

TMLR 2025

On Space Folds of ReLU Neural Networks

Michał Lewandowski, Hamid Eghbalzadeh, Bernhard Heinzl, Raphael Pisoni, and Bernhard A. Moser

In Transactions on Machine Learning Research, 2025

IJCAI 2022

Tessellation-Filtering ReLU Neural Networks

Bernhard A. Moser, Michał Lewandowski, Somayeh Kargaran, Werner Zellinger, Battista Biggio, and Christoph Koutschan

In International Joint Conference on Artificial Intelligence, 2022

BCP 2017

Geometric features of Vessiot-Guldberg Lie algebras of conformal and Killing vector fields on \(\mathbb{R}^2\)

Michał Lewandowski and Javier de Lucas

In Banach Center Publications, 2017

Workshop Papers (workshops are cool)

ICLR 2026 Workshop

A Graph-Theoretical View of Space Folding via the Motzkin–Straus Framework

Michał Lewandowski, Bernhard Heinzl, Roman Rainer, Bernhard Nessler, and Bernhard A. Moser

Accepted at multiple International Conference on Learning Representations workshops, 2026

ICLR 2025 Workshop

The Space Between: On Folding, Symmetries and Sampling

Michał Lewandowski, Bernhard Heinzl, Raphael Pisoni, and Bernhard A. Moser

Accepted at multiple International Conference on Learning Representations workshops, 2025

NeurIPS 2024 Workshop

CantorNet: A Sandbox for Testing Geometrical and Topological Complexity Measures

Michał Lewandowski, Hamid Eghbalzadeh, and Bernhard A. Moser

In NeurIPS Workshop on Symmetry and Geometry in Neural Representations, 2024

NeurIPS 2024 Workshop

The Turing Game

Michał Lewandowski, Simon Lucas Schmid, Patrick Mederitsch, Alexander Aufreiter, Gregor Aichinger, Felix Nessler, Severin Bergsmann, Viktor Szolga, Tobias Halmdienst, and Bernhard Nessler

In NeurIPS Workshop on System-2 Reasoning at Scale, 2024

Selected Applied Work (not under NDAs)

Surrogate Models

Machine-learning surrogates for rotary-kiln calcination

Details

At SCCH I work on machine-learning surrogates for calcination processes of limestone and magnesium carbonate in rotary kilns. The challenge is to approximate expensive physical simulations (CFD-based) with neural operator-based approaches. This simulation work is carried out within the PRIM-ROCK project.

Mechanistic Interpretability

Mechanistic interpretability of activation regions

Details

I work on questions around the semanticity of activation regions in LLMs and the geometry of CoT. The common theme is whether the information contained in the activation space can be used to mechanistically interpret model behavior.

Experience & Education

Employment

SCCH

Software Competence Center Hagenberg

Oct 2019 - present

Current role: Senior Researcher

  • Jul 2026 - present: Senior Researcher
  • Jan 2024 - Jun 2026: Researcher & Senior Data Scientist
  • Oct 2019 - Dec 2023: Researcher & Data Scientist
Inria

Research Intern at Inria

Grenoble, France · Feb 2018 – Aug 2018

Bayesian inference of extreme quantiles for my master’s thesis.

Education

PhD

PhD in Artificial Intelligence

Johannes Kepler University Linz · Feb 2022 - May 2025

(Undertaken alongside my employment at SCCH.)

MSc

MSc in Applied Mathematics

University of Warsaw / Inria Grenoble · Sep 2016 - Jul 2018

Overview
BSc

BSc in Mathematical Physics

University of Warsaw · Sep 2013 - Sep 2016

Overview

Further study

Timeline

Education Employment

Talks

Space Folding in Neural Networks

INTEGREAT Seminar, Tromsø, Norway, 2026.

On Space Folds by Neural Networks

ML in PL Conference, Warsaw, Poland, 2025.