Roman Arzumanyan
Team Leader at Okko
Moscow, Moscow, Russia
Summary
🤩
Rockstar🎓
Top SchoolRoman Arzumanyan is a hands-on team leader and senior video codec engineer with 8+ years building high-performance C++/CUDA systems for real-time video processing and analytics. He has led teams at Okko, VisionLabs and NVIDIA, delivering GPU-accelerated pipelines that process hundreds of video streams on a single GPU and bridging C++ cores to Python via pybind11. Roman combines low-level bit-fiddling and codec optimization (H.264/H.265) with solution architecture and customer-facing technical leadership, including work on NVIDIA's Video Processing Framework and FFmpeg integrations. A frequent OSS contributor, he has improved backend C++ libraries for NVIDIA's widely used VideoProcessingFramework, adding CUDA-aware conversions and debugging enhancements. He pairs academic rigour from a master's in applied mathematics with ongoing deep learning upskilling at Yandex Practicum, making him adept at shipping ML-enabled video features. Less obvious: he routinely turns research prototypes into production-ready modules and enforces release and branching policies that scale across multiple external customers.
8 years of coding experience
10 years of employment as a software developer
Yandex EdTechMaster's degree Applied Mathematics, Master's degree Applied Mathematics at Southern Federal University (former Rostov State University)Deep Learning, Deep Learning at Yandex Practicum
Английский, Русский
Stackoverflow
1,784reputation
123kreached
101answers
1question
Github Skills (12)
cuda10
c-language10
cprogramming-language10
ffmpeg9
kernel6
parallel-processing6
opencl6
buffer6
macos6
gpgpu6
gpu6
image6
Programming languages (5)
Github contributions (5)
NVIDIA/VideoProcessingFramework
Dec 2019 - Sep 2022
Set of Python bindings to C++ libraries which provides full HW acceleration for video decoding, encoding and GPU-accelerated color space and pixel format conversions
Role in this project:
Back-end Developer
Back-end DeveloperContributions:6 releases, 30 reviews, 353 commits in 2 years 9 months
Contributions summary:Roman contributed to the implementation and enhancement of the NVIDIA Video Processing Framework, primarily focusing on the backend C++ library. Their work included adding CUDA error messages for better debugging and providing support for YUV444-to-BGR color conversion and additional ffmpeg features. The commits also include changes to improve the handling of the decoder and encoder components, with optimizations and addition of new capabilities such as support for stream seeking.
pythonaccelerationvideoencoding-decodingcolor-space
rarzumanyan/HandBrake
Apr 2022 - Sep 2022
HandBrake's main development repository
Contributions:17 pushes in 5 months
javahandbrake
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.