Anton Vlasyuk
Senior Software Engineer at Banco Plata
Limassol, Cyprus
Summary
🤩
Rockstar🎓
Top SchoolAnton Vlasyuk is a Senior Java Developer based in Limassol, Cyprus with 4 years of professional experience designing and operating microservice-driven order and pricing systems for e-commerce and retail. He has architected and implemented robust order management and orchestration services using Kotlin/Java, Spring Boot, Camunda, Kafka, and Kubernetes, and previously built end-to-end pipelines that combine Android data collection, image ML preprocessing, and backend classification. Beyond backend systems, Anton contributes to high-profile open-source ML tooling—debugging and extending models in the Hugging Face Transformers repo, including GPT-2 and Mamba improvements—which reflects a rare bridge between production JVM services and ML model internals. Known for pragmatic, test-driven refactors and solving tricky caching and mask-generation bugs, he brings hands-on reliability to distributed systems and cross-functional ML integrations.
4 years of coding experience
13 years of employment as a software developer
Диплом специалиста по инженерной специальности Управление и информатика в технический системах, Диплом специалиста по инженерной специальности Управление и информатика в технический системах at МИРЭА
Github Skills (14)
flash10
pytorch10
transformer10
machine-learning10
multihead-attention10
gpt10
deep-learning10
self-attention10
mamba10
python10
natural-language-processing9
nlp9
testing8
jax5
Programming languages (3)
Github contributions (5)
huggingface/transformers
Mar 2024 - Jul 2026
🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.
Role in this project:
ML Engineer
ML EngineerContributions:789 reviews, 111 PRs, 320 pushes in 2 years 4 months
Contributions summary:Anton's primary contributions involve debugging and improving the Mamba and GPT2 models within the Hugging Face Transformers library. They fixed issues with the Mamba `slow_forward` gradient, implemented SDPA support for GPT2 and GPT-NeoX, and addressed problems related to caching and mask creation. Furthermore, they added new features and tests, and refactored existing code for greater efficiency.
audioinferencemachine-learning-modelsmultimodaltransformers
vasqu/mamba2-torch
Jun 2024 - Nov 2025
Contributions:1 review, 3 PRs, 32 pushes in 1 year 5 months
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.