Middle ML Engineer @ Sber · Kaggle Expert · Stepik streak: 648 days
I build applied ML systems across healthcare, computer vision, LLM tooling, IoT, and product engineering.
I like turning research ideas into working services, dashboards, and production-ready prototypes.
Languages & Data
ML & Infrastructure
Recent work I am especially excited about:
- Sintezum — university tech entrepreneurship platform with labs, organizations, vacancies, scientific requests, roles, admin flows, and production Docker setup.
- laser-energy-monitor — WPF desktop application for laser-energy measurement, stability checks, export, simulation mode, and BeamGage/Ophir equipment integrations.
- arduino-bridge — Arduino + ESP8266 bridge with a lightweight text protocol, FastAPI backend, PostgreSQL storage, nginx edge proxy, and device command queue.
- ai-copilot-for-learning-on-Stepik — product and architecture concept for a Stepik learning copilot that supports understanding without giving away ready answers.
- skin-cancer-ai + img2txt — interpretable dermoscopy pipeline with segmentation, Kittler-style decision logic, feature extraction, ranking, and clinical description generation.
| Project | Description | Stack | Live |
|---|---|---|---|
| Sintezum | Platform for university technology entrepreneurship: labs, organizations, vacancies, scientific requests, subscriptions, admin panel, and production ops docs | FastAPI React PostgreSQL Elasticsearch MinIO |
- |
| LDT | Real-time CTG monitor for early fetal hypoxia risk: live stream, FIGO-inspired features, CatBoost risk model, history mode, and clinical reports | Python FastAPI React ML Docker |
ctg-ldt.ru |
| skin-cancer-ai | Interpretable melanoma analysis system with YOLO/UNet segmentation, Kittler-style decision tree, Celery pipeline, auth, profile, API docs, and MinIO storage | Computer Vision FastAPI React Celery Docker |
skin-cancer-ai.ru |
| img2txt | Internal description service for dermoscopy cases: image+mask features, bucketing, importance ranking, LLM-based clinical text, and callback flow | Medical AI Computer Vision LLM SQLite |
skin-cancer-ai.ru |
| laser-energy-monitor | Desktop tool for laser-energy measurements, stability estimation, exports, simulation mode, and BeamGage/Ophir integrations | C# .NET Framework 4.8 WPF x86 |
- |
| arduino-bridge | IoT bridge between Arduino Uno/Nano + ESP8266 and a VPS backend with command/result/state/event protocol | Arduino ESP8266 FastAPI PostgreSQL nginx |
- |
| ai-copilot-for-learning-on-Stepik | Product docs and architecture for an AI sidebar that summarizes Stepik lessons, highlights pitfalls, and keeps learning guardrails explicit | Product LLM Browser Extension Architecture |
- |
| Sber Grace Assistant | Local finance assistant: SMS parsing via Ollama, grace-period tracking, Celery schedules, VK reports, investment lots, and CBR/MOEX integrations | Python Ollama FastAPI Celery PostgreSQL |
- |
| Practice | Early pneumonia diagnosis system based on chest X-ray images | Deep Learning Medical Imaging FastAPI |
api.pneumonia-ai.ru |
| SimpleAutoML | Lightweight AutoML system for computer vision experiments | AutoML CV Python |
- |
Now: Building scientific platforms, medical AI systems, hardware-adjacent tools, and practical LLM-assisted automation.
| Badge | Description |
|---|---|
| Kaggle Expert | Active participation, shared solutions and notebooks |
| Stepik | Daily learning streak: 648 days (auto-updated from profile) |
| Education | See Education section below |
- MEPhI — National Research Nuclear University
- Samsung Innovation Campus — educational track (AI / Big Data)
- Newprolab — DE Path (Data Engineering, Spark)
If you find any of my projects useful — a star would mean a lot.