Roman Vyunov
AI Software Engineer at YADRO
Nizhny Novgorod, Nizhny Novgorod Oblast, Russia
Summary
🤩
Rockstar🎓
Top SchoolRoman Vyunov is an AI software engineer with six years of experience building and optimizing neural network inference backends, currently focused on product development at YADRO. He is a Python-first developer skilled in FastAPI, SQLAlchemy, Pandas and low-level performance work, with deep hands-on experience in VPU backends through contributions to the widely used open-source OpenVINO toolkit. At Intel he worked on MLIR/OpenVINO-based inference frameworks, fixing customer model compilation issues, adding new layer and 3D support, and delivering platform-specific optimizations that improved real-world model performance. He also initiates tooling and debugging projects—most notably methods for analyzing restricted customer networks without topology access—demonstrating a pragmatic blend of research and delivery. Based in Nizhny Novgorod, Roman combines production-oriented engineering with deep inference-stack expertise that bridges high-level Python tooling and low-level accelerator nuances.
5 years of coding experience
3 years of employment as a software developer
Master's degree, Computer Software Engineering, Master's degree, Computer Software Engineering at State University of Nizhni Novgorod named after N.I. Lobachevsky (UNN)
Github Skills (10)
c-language10
deep-learning10
inference10
cprogramming-language10
back-end-development10
openvino10
optimization10
ml9
mle9
computer-vision9
Programming languages (1)
Github contributions (2)
openvinotoolkit/openvino
Jul 2020 - Feb 2021
OpenVINO™ is an open-source toolkit for optimizing and deploying AI inference
Role in this project:
Back-end & ML Engineer
Back-end & ML EngineerContributions:65 reviews, 10 commits, 22 PRs in 6 months
Contributions summary:Roman primarily contributed to the VPU (Vision Processing Unit) backend of the OpenVINO toolkit, focusing on optimizing AI inference. They implemented fixes for customer model compilation errors, specifically addressing issues related to data layout and constant inputs within the VPU. The user also refactored special stage processing and contributed to supporting new layer types and enabling 3D support for existing layers. They implemented new features like MVN and Swish layer support, integrating and adapting various neural network operations for the VPU.
inference-enginepytorchmodel-optimizerdeep-learninggpu
rvyunov/openvino
May 2020 - Nov 2021
OpenVINO™ Toolkit - Deep Learning Deployment Toolkit repository
Contributions:3 PRs, 71 pushes, 20 branches in 1 year 5 months
pytorchdeep-learningdeploymentopenvino-toolkitobject-detection
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.