Andrew Garkavyi
Engineering Manager at Grammarly
Berlin, Germany
Summary
👤
Senior🎓
Top SchoolAndrew Garkavyi is an engineering manager based in Berlin with over a decade of experience delivering award-winning consumer and embedded software products and leading cross-functional teams. He has driven measurable improvements in delivery velocity, quality, and retention—cutting release cycles in half and reducing post-release issues through CI/CD, automation, and product-focused planning. As former CEO of a product studio and now an EM at Grammarly, he blends hands-on technical depth (from low-latency Java EE systems to IoT and mobile) with strong program and people leadership. He has led large IoT deployments and mobile launches for luxury hospitality and consumer apps with millions of users, and mentors startups on product-market fit and engineering practices. Andrew also contributes to open-source NLP tooling—extending Stanford's Stanza NER to support transfer learning and Ukrainian datasets—reflecting his growing focus on AI/ML in physical-world applications. He combines entrepreneurial grit with a methodical approach to continuous improvement and pragmatic product thinking.
9 years of coding experience
12 years of employment as a software developer
Nano-degree, Natural Language Processin, Nano-degree, Natural Language Processin at UdacityIgor Sikorsky Kyiv Polytechnic Institute
English, German, Ukrainian
Github Skills (10)
pytorch10
machine-learning10
transfer-learning10
named-entity-recognition10
python10
natural-language-processing10
deeplearning-ai9
deep-learning9
data-pipeline8
data-pipelines8
Programming languages (4)
Github contributions (5)
stanfordnlp/stanza
May 2020 - Aug 2021
Stanford NLP Python library for tokenization, sentence segmentation, NER, and parsing of many human languages
Role in this project:
ML Engineer
ML EngineerContributions:5 commits, 2 PRs, 5 comments in 1 year 2 months
Contributions summary:Andrew primarily contributed to the Named Entity Recognition (NER) component of the Stanford NLP library. They implemented features to support transfer learning by allowing training of only the classifier layer. They also extended the NER tagger to support finetuning of existing model files and refactored the code to decouple the DataLoader and Trainer, allowing them to receive only the vocabulary and embeddings. Additionally, they added a conversion tool to transform the Ukrainian NER dataset into a compatible format for use with the library.
nlppytorchhuman-languagesnamed-entity-recognitionpython
gawy/stanza-lang-uk
May 2020 - Nov 2021
Converting training data set built by lang-uk community to the format supported by Stanza NLP library
Contributions:1 release, 14 commits, 2 PRs in 1 year 6 months
nlpstanford-nlpnamed-entity-recognitiondata-setlang
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.