Martin Sipka

Hello! I'm Martin Šípka, a researcher at Amazon Web Services working on Amazon Transcribe, where I focus on speech-to-text model development and their GPU-efficient implementations.

I obtained a PhD from Charles University under Michal Pavelka on the intersection of physics and machine learning, focusing mostly on enhanced sampling of chemical reactions. During my PhD, I was a visiting student at MIT, where I worked with Professor Rafael Gómez-Bombarelli on rare events sampling using differentiable molecular dynamics. My background includes strong expertise in C++, Python, and modern deep learning frameworks, along with experience in fluid dynamics and numerical mathematics.

My current research interests are focused on diffusion and flow matching models, particularly exploring the properties of energy landscapes that emerge during their training and the dynamical properties of their inference. Feel free to reach out if you'd like to discuss any of these topics!

I am a proud player on the Yetties hockey team in the Seattle Kraken Hockey League and an amateur guitarist looking for a spot in an equally amateur rock band.

Publications

Action-Minimization Meets Generative Modeling: Efficient Transition Path Sampling with the Onsager-Machlup Functional
Authors: S Raja*, M Šípka*, M Psenka*, T Kreiman, M Pavelka, AS Krishnapriyan
International Conference on Machine Learning, 2025
arXiv
A reactive neural network framework for water-loaded acidic zeolites
Authors: Erlebach, A., Šípka, M., Saha, I., Nachtigall, P., Heard, C.J., Grajciar, L.
Nature Communications 15(1), 4215, 2024
DOI
Differentiable Simulations for Enhanced Sampling of Rare Events
Authors: Sipka, M., Dietschreit, J.C.B., Grajciar, L., Gómez-Bombarelli, R.
International Conference on Machine Learning, 2023
DOI
Constructing Collective Variables Using Invariant Learned Representations
Authors: Šípka, M., Erlebach, A., Grajciar, L.
Journal of Chemical Theory and Computation 19(3), 887-901, 2023
DOI
Direct Poisson neural networks: learning non-symplectic mechanical systems
Authors: Šípka, M., Michal, P., Oğul, E., Miroslav, G.
Journal of Physics A: Mathematical and Theoretical 56(49), 2023
DOI
On compatibility of the natural configuration framework with general equation for non-equilibrium reversible–irreversible coupling (GENERIC): Derivation of anisotropic rate
Authors: Pelech, P., Tůma, K., Pavelka, M., Šípka, M., Sýkora, M.
Journal of Non-Newtonian Fluid Mechanics 305, 104808, 2022
DOI
Learning physics from data: a thermodynamic interpretation
Authors: Chinesta, F., Cueto, E., Grmela, M., Moya, B., Pavelka, M., Šípka, M.
Workshop on Joint Structures and Common Foundations of Statistical Physics, 2020
DOI
Lack-of-fit reduction in non-equilibrium thermodynamics applied to the Kac–Zwanzig model
Authors: Mladá, K., Šípka, M., Pavelka, M.
Journal of Non-Equilibrium Thermodynamics 49(2), 181-194, 2024
DOI