Grigory Malivenko
Machine Learning Engineer at Pixelmator Team
Osaka, Japan
Summary
👤
Senior🎓
Top SchoolGrigory Malivenko is a machine learning researcher and engineer with 11 years of experience and a deep focus on computer vision, medical imaging, and NLP, currently based in Osaka and working at AI Witchlabs AG. He has built production-ready image-to-image models (super-resolution, denoising, color enhancement) at Pixelmator and delivered contract computer vision work for Apple, combining research rigor with product-grade deployment. Comfortable across the stack, he integrates ML into constrained environments using C/C++ and optimizes models for edge and mobile runtimes (TensorFlow Lite/LiteRT, CoreML, ONNX, TensorFlow.js). He contributes to open-source tooling—authoring a PyTorch-to-Keras converter that handles core layers and architectures—reflecting a practical bent for interoperability and fast prototyping. Ongoing personal research includes NLP for resource-limited language pairs (RU-LT, EN-LT) and compact CV models like PyNet-V2 Mobile, showing a sustained interest in efficient ML for real-world constraints.
11 years of coding experience
9 years of employment as a software developer
Master's degree Software Engineering, Master's degree Software Engineering at Southern Federal University (former Rostov State University)Bachelor's degree Mathematics and Computer Science, Bachelor's degree Mathematics and Computer Science at ICTIS Southern Federal University (former Rostov State University)
English, Russian
Github Skills (9)
neural-network10
keras10
pytorch10
machine-learning10
convolutional-neural-networks10
model-conversion10
deep-learning10
resnet10
python10
Programming languages (8)
Github contributions (5)
gmalivenko/pytorch2keras
Nov 2017 - Aug 2021
PyTorch to Keras model convertor
Role in this project:
ML Engineer
ML EngineerContributions:1 release, 255 commits, 14 PRs in 3 years 9 months
Contributions summary:Grigory primarily contributed to the development of a PyTorch-to-Keras model converter. Their work involved creating the project structure and implementing conversion logic for various PyTorch layers such as Dense, Convolutional, Batch Normalization, Pooling, and activation layers (ReLU, LeakyReLU, etc.). They also added tests for the implemented layers and included support for the ResNet and other model architectures.
mobilenetmlptensorflowjskeras-modelsdarts
gmalivenko/gluon2pytorch
Oct 2018 - May 2021
Gluon to PyTorch deep neural network model converter
Contributions:130 commits, 1 PR, 62 pushes in 2 years 6 months
pytorchdeep-learningdeep-neural-networkneural-networkgluon
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.