Работал в 2 компаниях 2 года 2 месяца
IT
ML/Robotics Engineer
NDA
IT
4 месяца
05.2026 - по настоящее время
Software Engineer
Skynav AI
IT
1 год 10 месяцев
07.2024 - 04.2026
- Machine Learning: Designed and trained custom CNN models using TensorFlow, Keras, and TensorBoard, incorporating convolutional, pooling, dropout, flatten, and dense layers; executed training on RTX GPU servers via SSH while exploring deep learning, supervised learning, reinforcement learning, and evaluation metrics.
- Data Preparation for Machine Learning: Built proprietary HDF5 datasets from UAV video footage (with embedded coordinate and camera-orientation logs) and simulated environments in Unreal Engine with Cesium; applied data augmentation, normalization, train/validation/test splitting, and OpenCV preprocessing to produce high-quality training data.
- Navigation System: Developed a proprietary computer vision-based navigation system for UAVs, utilizing onboard CNN models for real-time image descriptor generation and efficient vector similarity search against offline map embeddings with FAISS and KD-Tree, including targeted optimizations for memory management and performance on embedded platforms.
- Embedded Programming: Implemented Python-based embedded solutions integrating MAVLink and ArduPilot protocols; optimized navigation deployment across edge devices (Raspberry Pi, NVIDIA Jetson) and multi-GPU UAV platforms through tensor operations, multiprocessing, multithreading, and performance/memory tuning.
- API Development: Architected and built the complete Tiles Mission Planner – an end-to-end automated drone photogrammetry system using FastAPI and Docker. Designed the full pipeline covering deterministic mission planning (waypoint grid generation with standard photogrammetry overlaps), tile processing (NodeODM integration to generate georeferenced orthomosaics from raw UAV imagery), and a stateless RESTful XYZ Tile API server. Delivered production-ready web map tiles with layer versioning, transparent-tile fallback, coverage queries, and seamless integration for simulation environments such as Unreal Engine.
- Multi-Agent Systems: Architected a custom multi-agent orchestration framework using prompt engineering and communication protocols to enable autonomous CNN model training without external dependencies.
Ключевая информация
- Python
- Machine Learning
- Computer Vision
- Edge Devices
- Linux
- Bash
- Git
- Prompt Engineering
- RAG
- AI Tools
- Python
- Git
- Machine Learning
- Вміння працювати в команді
Учился в 1 заведении
Київський національний університет імені Тараса Шевченка
Факультет інформаційних технологій
2025
Владеет языками
Английский
продвинутый
Может проходить собеседование на этом языке
Может проходить собеседование на этом языке
Русский
родной
Может проходить собеседование на этом языке
Может проходить собеседование на этом языке
Украинский
родной
Может проходить собеседование на этом языке
Может проходить собеседование на этом языке
Курсы, тренинги, сертификаты
SOFTSERVE CERTIFICATE
Vladyslav Lekomtsev, member of theCodeGalacticos team, which took first place in theSPU, SPFIT and SOFTSERVE hackathon.
Vlad Mykhailovich
Vlad Mykhailovich
Software Engineer/Computer Vision Engineer/Machine Learning Engineer
Киев
Готов переехать: Львов, Одесса, Днепр
Активно ищет работу
полная занятость, неполная занятость, проектная работа
Характер работы: удаленная работа
Последняя активность 15 часов назад