Cto
Current
Saint-Petersburg, Russia
Started the inner department of single-board (SoC and SoM) Ai and neural nets solutions.Management of the software development department: full-stack (low-level hardware programming, backend, decentralized core, neural network algorithms for non-standard architectures and their adaptation, logical wrappers for business functions, backend web development, UI / UX), developed cloud flexible web service for a zoo of SoM Ai devices.Designing software architecture and managing the development process of the project for the creation of an automotive system with face identification, tracking behavior, and checking the ethanol vapor content of an individual.Release of low-cost commercial products based on single-board computers in a short development cycle. Managing tasks from prototypes of commercial hypotheses, choosing a hardware platform to develop custom algorithms and UI / UX debugging of interfaces. Constant practical research of modern realities of the industry, scientific articles, frameworks.Working with hardware platforms:-Nvidia: Jetson nano, Jetson Xavier AGX, Jetson Xavier Tx2,-Google: Coral dev board, USB EdgeTPU accelerator -Raspberry Pi 3, 4 - Khadas Vim3- Custom board with NXP 8m+ (Cortex A53 with NPU) Core.- Intel MovidiusLinux working software tool stack: Ubuntu, Mendel, CentOS; Neural: TensorFlow, TFLite, ONNX, OpenCV, dlib, Pytorch, Detectron2, TensorRT, Nvidia DLA, media pipe. Dev: Python, C, C ++, bash, GStreamer, FFmpeg, SQL, etcBuilding a public cloud service for a network of neural network analytical endpoints. Leading the development processes for creating proprietary neural network algorithms: FaceID, determining the angle of head rotation, closing the eyes, crossing the line, counting objects in the area, detecting mood, age, etc...Working with B2B clients, building a strategy, software architectures and development plans, full technical management of several projects related to the neural networks, and complex developments.
