07/26/2026 updated
100 % available
Computer Vision Engineer | Object Detection & Video Analytics | 10 yrs
Bratislava, Slovakia
Only remote
M.Sc. Mathematics, Taras Shevchenko National University of KyivAbout me
Computer Vision engineer with 10 years in production ML. I build object detection and video analytics systems that hold up on live camera streams, not just on benchmarks. Available part-time or full-time, fully remote. Working languages: English, Russian, Ukrainian.
Artificial IntelligenceComputer VisionLinuxImage ProcessingPython (Programming Language)Machine LearningObject DetectionOpenCVTensorFlowVisual AnalyticsData SciencePyTorchDeep LearningScikit LearnMachine Learning Operations
10 years of commercial experience in computer vision and machine learning (since 2016), focused on object detection and real-time video analytics in production. Available for part-time or full-time engagements, fully remote, English-speaking projects.
Computer Vision
Expertise in real-time object detection using RT-DETR and YOLO family models for tasks such as vessel detection, security-relevant object detection, person-overboard detection, and fraudulent behavior recognition. Proficiency in segmentation, OCR, monocular depth estimation, and applying vision-language models (Qwen-VL, Gemma 3, SmolVLM2) for scene understanding.
Deep Learning & Model Development
Hands-on experience with PyTorch, PyTorch Lightning, and TensorFlow for building and training deep learning models. Development of custom training frameworks, CNNs for object detection and tracking, and model architectures optimized for edge inference on resource-constrained devices.
MLOps & Deployment
Model training and optimization for edge inference, API deployment, containerization with Docker on Linux, and version control with Git. Experience deploying ML models in production environments across various domains, including continuous video streams from CCTV cameras.
NLP / LLM & RAG Systems
Development of Retrieval-Augmented Generation (RAG) systems for LLM-powered product search and customer support in e-commerce. Experience with vector databases, LLM-powered search and assistants, and OCR systems for document field detection and localization.
Multimodal / Vision-Language Models
Application of multimodal models such as Qwen-VL, Gemma 3, and SmolVLM2 for auxiliary detection tasks and scene understanding in complex environments.
Classical Machine Learning
Proficiency in scikit-learn for regression, classification, clustering, and time-series forecasting. End-to-end ML pipelines on tabular data including data preparation, metric selection, model training, and API deployment.
Tabular ML & Financial Analytics
Experience with sales and revenue forecasting from historical data, food demand trend forecasting, and customer credit forecasting based on client profile evaluation using nationwide financial data.
Programming Languages
Python with 10+ years of experience as the primary language for ML and data science projects. Additional experience with C++.
Experience
Professional experience10+ years
Freelancer since2018
Languages
GermanBasic knowledgeEnglishFluentRussianNative speakerUkrainianNative speaker
Project history
Computer vision for maritime security and video surveillance systems.
- Training and optimization of real-time object detection models (RT-DETR, YOLO family) on continuous CCTV streams
- Detection classes include vessels, weapons and other security-relevant objects, person-overboard events, and anomalous behaviour
- Handling long-range small-object detection, low-light and adverse weather conditions, and class imbalance for rare events
- Design of evaluation protocols reflecting real operating conditions rather than benchmark scores
- Model optimization for edge inference on resource-constrained devices
- Deployment in production: containerization with Docker on Linux, API services, version control with Git
Developed an OCR system and trained detectors for document field detection and localization. Built a Retrieval-Augmented Generation (RAG) system for LLM-powered product search and customer support in e-commerce. Developed ML models for real estate price prediction based on property features and market data. Built a computer vision system to detect fraudulent behavior at self-checkout stations in large retail environments.
Indoor scene segmentation and generative scape planning for interior design applications. Built a custom deep learning training framework on top of PyTorch Lightning. Sales and revenue forecasting from historical sales data and contextual features. Designed and trained a sound classification algorithm.