Rafał Lewandków

Computer Vision / Scientific Imaging / Research Software

rafal.lewandkow2@uwr.edu.pl | r.lewandkow@gmail.com | ORCID: 0000-0001-9683-1084 | GitHub: github.com/RafLew84

EDUCATION

Doctor of Physical Sciences - University of Wroclaw | 2023

Master of Physics - University of Wroclaw | 2017

PROFILE

Experimental physicist and research software developer working at the intersection of scientific imaging, computer vision and applied AI. I design tools that translate domain-specific problems into usable workflows. I specialize in analysing imperfect STM/EC-STM and LEED image data: low contrast, drift, instrumental artefacts, small datasets and correlated frames. I combine classical image processing with deep learning when it meaningfully improves detection, tracking, segmentation, quantitative validation or human-in-the-loop curation.

SKILLS

Python engineering

Python, NumPy, SciPy, pandas, PyQt6, pyqtgraph, matplotlib, Jupyter, pytest, Git.

Computer vision / ML

PyTorch, Ultralytics YOLO, U-Net, SAM/SAM2, OpenCV, scikit-image; detection, segmentation, tracking, optical flow, registration.

Scientific imaging

STM, EC-STM, LEED, XPS, UPS; noisy data, instrumental artefacts, analysis of interfaces and surface structures.

Validation

YOLO-format dataset building, train/val/test split for correlated frames, annotation QC, benchmarks, documentation.

EXPERIENCE

Assistant Professor - University of Wroclaw

2024-present

Development of ML/CV tools for STM/EC-STM and LEED analysis: object detection, segmentation, tracking, registration and quantitative metrology.

Training and integration of YOLO detectors for STM/EC-STM molecules and LEED spots; using ROI-based detection, tracking/linking and downstream analysis.

Design of human-in-the-loop workflows: manual annotation correction, sessions, data export, geometric validation and user documentation.

Teaching and preparation of materials for mobile-device programming and deep learning courses.

Lecturer - University of Wroclaw

2021-2024

Teaching and preparation of programming and project materials for students.

Teaching Assistant - University of Wroclaw

2018-2021

Research and teaching work in experimental physics, surfaces and interfaces.

Analyst - BM@N Collaboration, JINR Dubna

2020-2022

Preparation of equation-of-state (EoS) tables for the THESEUS code in the NICA/BM@N programme; work with research code and reproducible data-processing workflows.

SELECTED PROJECTS AND TOOLS

MolDetA - PyQt6 platform for curation and analysis of multichannel STM/EC-STM data: registration as a mapping layer, YOLO/classical detection, bbox propagation, manual correction, grouping, sessions and export.

LFA - Published open-source Python/PyQt6 tool for quantitative STM/EC-STM metrology: preprocessing, FFT, peak localization, affine correction, lattice parameters, uncertainties and sessions.

STAL - Workbench for LEED spot detection, tracking and analysis: classical processing, YOLO, Hungarian linking, optical flow. Integration of a tracker architecture based on a registry of point, bounding-box and mask/video backends.

NaParA / NanoTrack / QNA - Tools for analysing nanoparticles, STM sequences, step-edge tracking and ROI-based quantum noise analysis, with emphasis on GUI, validation and reproducible exports.

SELECTED PUBLICATIONS

Lewandków, R.; Wira, P.; Futyma, A.; Wasielewski, R.; Kosmala, T. LFA: A Lattice Fourier Analyzer for Quantitative In Situ EC-STM of Adsorbate-Substrate Superstructures. Advanced Materials Interfaces, 2026. DOI: 10.1002/admi.70500.

Grodzicki, M.; Sabik, A.; Mazur, P.; Tołłoczko, A. K.; Lewandków, R. et al. Band alignment of amorphous Ge2S3 and GaN(0001). Journal of Materials Science, 2026.

Lewandków, R.; Mazur, P.; Grodzicki, M. Niobium oxides films on GaN: Photoelectron spectroscopy study. Thin Solid Films, 2022, 763, 139573. DOI: 10.1016/j.tsf.2022.139573.

Lewandków, R.; Grodzicki, M.; Mazur, P.; Ciszewski, A. Interface formation of Al2O3 on carbon enriched 6H-SiC(0001): Photoelectron spectroscopy studies. Vacuum, 2020, 177, 109345. DOI: 10.1016/j.vacuum.2020.109345.