_______________________________
Hey, I work as a Senior Data Scientist in the Machine Learning for Verification group at Arm. Previously, I was a Research Software Engineer in the Drosophila Connectomics lab at University of Cambridge. I'm interested in machine learning, neuroscience and text processing. As a Software Sustainability Institute Fellow, I advocate for open science, by promoting good coding practices in research and by contributing to open source software.
In the past I worked on: connectomics, brain computer-interfaces, psychophysics, brain functional connectivity, natural language processing, R packages and artificial neural networks.
Aya Dataset: An Open-Access Collection for Multilingual Instruction Tuning | Singh S, ..., Krzemiński D, ..., Hooker S (2024) ACL 2024
Open source software for automatic subregional assessment of knee cartilage degradation using quantitative T2 relaxometry and deep learning | Thomas KA, Krzemiński D, Kidziński Ł, Paul R, Rubin EB, Halilaj E, Black MS, Chaudhari A, Gold GE, Delp SL (2021) Cartilage
Energy landscape of resting magnetoencephalography reveals fronto-parietal network impairments in epilepsy | Krzemiński D, Masuda N, Hamandi K, Singh KD, Routley B, Zhang J (2020) Network Neuroscience
Helmholtz principle on word embeddings for automatic document segmentation | Krzemiński D, Balinsky H, Balinsky A (2018) 18th ACM Symposium on Document Engineering
Breakdown of long-range temporal correlations in brain oscillations during general anesthesia | Krzemiński D, Kamiński M, Marchewka A, Bola M (2017) Neuroimage
Not so artificial neurons [PL] - article in Polish popular science magazine Delta; 2018
Internationalization of shiny apps has never been easier - 2018
TODOr - addin to RStudio, which helps discover all TODO places in the RStudio (addin of the day by RStudio).
Websites for Health Assesment Tool - provides users with a ranking of online medical information sources credibility. Created at NHS HackDay 2020 - 3rd place in a community vote. Chrome extension available to download from the Chrome store. Also, see sister project: AuthentiSci.
ARmadillo - augmented reality 3D MRI data visualization; created at OHBM Hackathon 2018 in Singapore.
DFA python - an implementation of Detrended Fluctuation Analysis algorithm.
ConnectiviPy - Google Summer of Code 2016 project. From scratch implementation of a Python module with various brain connectivity estimators.
niftytorchprep - CLI for preparing neuroimaging data from BIDS format to train/test/validation structure for deep-learning training in NiftyTorch.
Contributions to open source packages: sklearn, shiny.semantic, shiny.collections, shiny.i18n, semantic.dashboard, mljar, brian2, mne-python and more...
Non-exhaustive list of my public presentations.
Conferences:
SfN 2022 (San Diego), PyConIT 2022 (Florence), Connectomics 2022 (Berlin), SBDM 2021 (Paris), MLPrague 2021 (online), MLinPL'20 (online), Neuromatch2 (online), SfN 2019 (Chicago), ITAI 2019 (Cardiff), HEALTAC 2019 (Cardiff), MEG UK 2019 (Cardiff), Science Polish Perspectives 2018 (Oxford), OHBM 2018 (Singapore), MEG UK 2018 (Londonderry/Derry), Brainstorm 2016 (Warsaw), Mind, Brain and Body Symposium 2016 (Berlin), Neuronus 2015 (Cracow), 4th Aspects of Neuroscience 2014 (Warsaw)
Workshops:
Code Review Workshop 2022, Big Data and High-Performance Computing 2018 & 2019 (invited talk), Shiny semantic workshop at SatRday Gdańsk 2019, 4th Baltic-Nordic Summer School on Neuroinformatics 2016
Meet-ups:
SatRday Amsterdam, PyData Cardiff, CaRdiff
I am/was active member of various tech communities:
(2023-24) European Regional Lead for Aya project by Cohere For AI.
(2021-23) Data Champion at University of Cambridge.
(2020) Tutoring and co-organising Brain Modelling Workshop at Cardiff University;
(2020) Tutoring Shiny workshop, organised by CaRdiff R Users group;
(2019) Co-organising SatRday Cardiff;
(2019) Helping out at BMVC conference in Cardiff;
(2018) Organising workshop Best programming practices for open science for GW4. The workshop aimed to promote good programming standards for reproducible research.