Vladimir Luchinskiy
Computer Vision Engineer at ATOM
Moscow, Moscow, Russia
Summary
👤
Senior🎓
Top SchoolVladimir Luchinskiy is a Computer Vision Engineer with eight years of experience and nearly three years focused on deep learning, specializing in model quantization, NPU deployment, and productionizing vision pipelines. He has driven end-to-end efforts from synthetic data generation and OCR pipelines to face recognition and anti-spoofing systems, consistently converting PyTorch models to optimized runtime formats like TensorRT and ONNX for embedded and NPU targets. At ATOM he built a library for automatic ONNX modification and quantization workflows and now reviews architectures for NPU execution, demonstrating a rare combination of research rigor and production engineering. His background includes robotics perception, GAN-based data augmentation that materially improved winter-scene accuracy, and GUI tools for automated labeling—skills that bridge data, model, and deployment gaps. Comfortable writing both Python and C++ inference integrations and lightweight custom layers for quantized runtimes, he excels at squeezing high accuracy from constrained hardware. Based in Moscow, he pairs a mathematical systems education with practical wins in product-facing computer vision.
8 years of coding experience
5 years of employment as a software developer
Степень бакалавра, Математическое обеспечение и администрирование информационных систем, Степень бакалавра, Математическое обеспечение и администрирование информационных систем at Санкт-Петербургский Государственный Университет
Github Skills (22)
segmentation6
semantic-segmentation6
pytorch5
face4
deep-learning4
semantic4
computer-vision4
face-recognition3
face-detection3
tensorflow3
cpp3
trading3
inference3
autoencoder2
onnx2
Programming languages (3)
Github contributions (5)
ucLh/city_obj_retrieval
Mar 2020 - Jun 2020
C++ module for inferencing tensorflow convolutional networks
Contributions:41 commits, 6 PRs, 20 pushes in 3 months
cppc-moduleconvolutionalconvolutional-networkstensorflow
ucLh/torch2onnx2trt
Sep 2021 - Feb 2022
A simple package that wraps PyTorch models conversion to ONNX and TensorRT
Contributions:39 commits, 16 PRs, 33 pushes in 5 months
pytorchpytorch-modelsdeep-learningwrapsonnx
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.