Hi, I'm Adam.
I am
Eight years building NLP systems across academia and industry — from yeast cell tracking to multilingual NER and ASR rescoring. I co-organize PolEval and contribute to CLARIN-PL open-source tooling.
Services
Engagements range from short-form research sprints to long-term embedded engineering.
Deep Learning Engineering
End-to-end design and training of neural architectures — from data pipelines to production deployment on GPU clusters.
NLP Systems
Custom language models, NER, coreference, ASR rescoring and information extraction across low-resource languages.
Research & Consulting
Translating academic SoTA into shippable systems. Literature reviews, prototypes, and reproducible benchmarks.
Open-source Tooling
Building and contributing to NLP toolkits and annotation platforms used by research communities.
Research Projects
A decade of applied research across NLP, computer vision, and open-source tooling.
BSNLP 2021 — NER & Linking
↗Few-shot and zero-shot Named Entity Recognition and coreference resolution across multilingual Slavic languages (Polish, Czech, Ukrainian).
PolEval 2021 — Task 1
↗Co-authored the punctuation restoration task: data acquisition, annotation guidelines, evaluation metrics, and contestant scoring.
PolEval 2020 — t-REx for ASR
↗Hybrid ASR improvement system: lattice extender + transformer-based utterance rescorer using ELECTRA.
BSNLP 2019 — Slavic NER
↗Shared task on multilingual NER and entity disambiguation across Slavic languages.
PolEval 2018 — Nested NER
↗Led a team on Nested Named Entity Recognition for Polish, proposing SoTA methods and final submission.
Liner2 & Crete Coreference
↗Open-source NER and coreference resolution toolkit developed at CLARIN-PL.
Inforex Annotation Tool
↗Web system for collaborative text corpora construction with multi-level semantic annotation.
CellStar Algorithm
↗Automatic segmentation and tracking of budding yeast cells in brightfield time-lapse microscopy.
Yeast Image Toolkit
↗Benchmarking platform for cell segmentation and tracking algorithms in microscopy.
AudioScope
Research system for identifying spoken phrases in audio recordings.
Skills & Certificates
- PhD candidate — Wrocław University of Science and Technologyongoing
- MSc — Computer Science, WUST2017
- PolEval Co-organizer2018–2021
- BSNLP Shared Task contributor2019, 2021
Publications
- 01↗Slavic Named Entity Recognition: BSNLP 2021 Shared Task SubmissionBSNLP @ EACL 2021
- 02↗PolEval 2021 Task 1: Punctuation Restoration from Read TextPolEval 2021
- 03↗t-REx: Transformer Rescorer & Extender for ASRPolEval 2020
- 04↗Nested Named Entity Recognition for PolishPolEval 2018
- 05↗CellStar: Algorithm for Yeast Cell Segmentation in Brightfield MicroscopyBioinformatics
Blog posts
Occasional notes from the lab bench and the production trenches.
Notes on training ELECTRA from scratch in Polish
Lessons learned from pre-training a Polish ELECTRA model on a single 8×V100 node — data curation, masking strategies, and gotchas.
Few-shot NER across Slavic languages
How meta-learning and cross-lingual transfer can salvage NER quality when annotated data is essentially absent.
Building reproducible NLP benchmarks
What three years of co-organizing PolEval taught me about evaluation design, data leakage, and contestant-friendly tooling.