In 2010 I graduated with an MSc in Engineering from the Electronics and Telecommunications program at Poznań University of Technology, multimedia specialization, master's thesis on synthesis of musical instrument sounds. In 2025 I defended my PhD in Computer Science at the Faculty of Mathematics and Computer Science at Adam Mickiewicz University in Poznań (defense presentation). My research focused on evaluation of speech recognition systems, culminating in a publication at NeurIPS and an open challenge at PolEval.
I currently collaborate with the Faculty of Mathematics and Computer Science at AMU, teaching a workshop on ASR systems evaluation (syllabus) and taking part in R&D projects on audio processing at the Center for Artificial Intelligence.
Professionally, I specialise in leading interdisciplinary R&D teams and managing networks of external partners, with a track record of delivering data products and services effectively and on time. I am currently the lead of the Data Annotation & QA Services team at Allegro, responsible for infrastructure and quality assurance processes for AI solutions in Poland and international markets. Full details in my CV.
Outside work I popularise AI systems evaluation at ekspertwpetli.pl, write about children’s digital safety at tatai.pl, make electronic music as Me?How? (BC, SC, YT) and spend time with my family.
Talks
- 2026GenAI in Localization 2026
- 2026GenAI in Localization 2026
Run with Allegro's Localization & Data team.
- 2025Product Cafe #15
- 2025GenAI in Localization 2025
- 2024NeurIPS 2024 (D&B Track)
- 2024PolEval 2024
- 2024YouTube
- 2024Panel discussion (YouTube)
Artifacts
- ASR TCO Calculator — Indicative TCO for Polish speech-to-text — API, transcription correction, complaint handling. Powered by BIGOS data.
- BIGOS v2 (HuggingFace) — Curated benchmark grouping 24 open speech corpora (BIGOS V2 and PELCRA for BIGOS) for reproducible Polish ASR evaluation. Published at NeurIPS 2024.
- Polish ASR Leaderboard — 25 ASR systems benchmarked live on 4000+ Polish recordings — accuracy, cost, speed, robustness.
Recommended reading
- Designing Machine Learning Systems — Chip Huyen
Najlepsze pojedyncze źródło o ML w produkcji jakie znam.
- Speech and Language Processing — Jurafsky & Martin
Klasyk; obecnie 3rd edition draft.
- Niespokojne pokolenie — Jonathan Haidtcurrently reading
Empirycznie ugruntowana krytyka smartfonowego dzieciństwa.
Contact
Open to academic collaboration, BIGOS questions, voice and speech evaluation discussions.
