Oleksandr Skurzhanskyi
Technical Lead, ML at Grammarly
Berlin, Germany
Summary
👤
Senior🎓
Top SchoolOleksandr Skurzhanskyi is a Technical Lead in ML at Grammarly with a decade of experience building and shipping applied machine learning systems, currently leading a team of 10+ on LLMs, instruction tuning, preference optimization and quality evaluation. He combines deep mathematical training (BSc/MSc with honors, PhD studies in AI) with hands-on engineering, having delivered state-of-the-art non-autoregressive GEC models and an on-device GEC solution that preserved 99% of quality under strict ROM/RAM constraints. His work spans research and production: from novel evaluation metrics for text generation to optimizing tokenization, post-processing and CPU inference in open-source projects like grammarly/gector. At LUN he built end-to-end ML pipelines for geo-recognition, duplicate detection and CTR improvements, showing a rare mix of signal-processing intuition and product impact. Based in Berlin, he is particularly interested in multimodal generative AI and pragmatic model compression strategies that make research deployable. Colleagues value him for translating academic rigor into production-ready, interpretable NLP systems.
10 years of coding experience
5 years of employment as a software developer
Doctor of Philosophy - PhD, Computer Science, Artificial Intelligence, Doctor of Philosophy - PhD, Computer Science, Artificial Intelligence at Taras Shevchenko National University of Kyiv
English, Ukrainian, German
Stackoverflow
21reputation
2kreached
0answers
1question
Github Skills (10)
transformers10
pytorch10
nlp10
python10
natural-language-processing10
bert10
roberta9
sequence-labeling9
measurement-protocol6
google-analytics6
Programming languages (5)
Github contributions (5)
grammarly/gector
May 2020 - Oct 2022
Official implementation of the papers "GECToR – Grammatical Error Correction: Tag, Not Rewrite" (BEA-20) and "Text Simplification by Tagging" (BEA-21)
Role in this project:
ML Engineer
ML EngineerContributions:24 commits, 10 PRs, 12 pushes in 2 years 5 months
Contributions summary:Oleksandr primarily contributed to the implementation and improvement of a grammatical error correction (GEC) model, as well as text simplification tasks, within the repository. Their work included removing duplicate variables, optimizing the model to run on CPU, fixing max length issues during post-processing, and fixing tokenization and training parameters. Additionally, they added confidence values and filter brackets, along with normalization for text simplification, demonstrating a focus on improving model performance.
grammatical-error-correctionnlptransformersxlnetbert
VikaMartynova/TodoApp
Nov 2018 - Nov 2018
Contributions:7 pushes in 1 day
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.