Description

Hey there! 👋

I’m a Computer Vision Engineer with 7 years of experience developing and deploying high-performance vision systems for real-world use cases. My focus lies in scalable deep learning, robust data workflows, and efficient deployment for both cloud and resource-constrained edge environments.

CORE SKILLS

Training: PyTorch + Lightning, RayTune-based distributed tuning, CI/CD, containerized experimentation (Docker, docker-compose)

Advanced ML: adversarial robustness (rFGSM, rPGD), neural architecture search (NAS), domain generalization, domain shift mitigation

Data Inspection: FiftyOne (Voxel51), embedding-based visualization, matplotlib, seaborn

Data Workflows: scalable CVAT-based pipelines, model-in-the-loop automation, reproducible data versioning (DVC, MLflow), synthetic data generation (CLIP, StableDiffusion)

Optimization: quantization, pruning, knowledge distillation, model compression

Deployment: Triton Inference Server (cloud and on-prem), GPU preprocessing with DALI, DeepStream & TensorRT (Jetson), OpenVINO, custom GStreamer plugins, CUDA-accelerated inference, real-time C++/OpenCV systems

Industry field of expertise

Languages

Russian
Native or bilingual
English
Fluent
German
Fluent

Workplace preferences

Can work on-site

Berlin (up to 50km)

brighter AI
Senior Deep Learning Engineer
SOFTWARE PUBLISHING
October 2024 - Today (1 year and 8 months)
Berlin, Germany
Leading the design and delivery of production-grade, real-time video analytics and edge AI systems. Currently driving model optimization and deployment strategies using TensorRT, DeepStream, and custom GStreamer pipelines (C++), ensuring scalable and reliable performance across cloud and edge environments.
GStreamer DeepStream Technical Leadership Edge AI Production ML
brighter AI
Deep Learning Engineer
DIGITAL AND IT
November 2022 - October 2024 (1 year and 11 months)
Berlin, Germany
Worked on developing, training, and deploying computer vision models for production use, covering the full ML lifecycle from data preparation to optimized inference. Focused on improving model accuracy and robustness using PyTorch, adversarial training techniques, and systematic data inspection. Built scalable experimentation and CI/CD workflows, and deployed optimized models to cloud and edge environments using TensorRT, Triton Inference Server, and DeepStream.
Pytorch TensorRT CI/CD ML lifecycle Data management
Scientific Research Institute of Radioelectronics
Computer Vision Engineer
DIGITAL AND IT
June 2019 - October 2022 (3 years and 4 months)
Worked on research-driven development of computer vision models for industrial applications. Focused on designing and evaluating detection and segmentation approaches, and later deploying research results into real-time systems using GPU-accelerated pipelines and C++-based video processing.
OpenCV Jetson Research