ML Engineer, KP Labs
Silesian University of Technology

Jakub Jaromin

ML researcher working on physics-informed and space AI

A neural network flowing into a wave surface

Currently

Invertible Multiscale Transforms on Manifolds

Learnable, exactly invertible multiscale transforms for manifold-valued signals.

GNOME anomaly detection

Anomaly detection in optical-magnetometer data for dark-matter searches.

Bachelor of Engineering Thesis

PINNs solving the Einstein equations for the Kerr metric.

Selected publications

  1. Data augmentation through angular–radial segmentation

    Jakub Jaromin, Marcin Woźniak

    Knowledge-Based Systems · Available online 31 July 2026 · published

    Ongoing: generalisation of the segmentation to n dimensions via pairs of PCA components.

  2. Playing Card Recognition Using Deep Neural Networks: Architecture Design and Performance Analysis

    Jakub Jaromin

    IVUS 2025, CEUR Workshop Proceedings, Vol. 4213 · 2026 · pp. 145–151 · published

All publications, posters and work in progress →

Awards

2025

HackYeah 2025 — 1st place, Biohacking

Rest & Blink app; ~60 teams in the category.

2025

Grant „Najlepsi z najlepszych! 4.0”

Ministry of Science and Higher Education.

2025

SGH × Mastercard Hackathon 2025 — finalist

Top 8 of 140 teams.

Contact

Open to research collaborations, R&D work in space and physics-informed ML.

Write an email