English · Русский
Building self-hosted AI assistants, monitoring systems, ML pipelines and developer tools.
- 🔭 Building local and privacy-focused RAG applications
- 🧠 Interested in LLM infrastructure, MLOps and AI observability
- ⚙️ Developing backend services with Python, Go and PHP
- 🤖 Shipping MAX messenger bots and the tooling around them
- 📊 Creating monitoring tools with Prometheus and Grafana
- 🐳 Packaging reproducible environments with Docker and CI/CD
- 🤝 Open to backend, AI infrastructure and open-source collaboration
- Local LLM applications powered by Ollama
- MAX messenger bots: templates, CI/CD and a GitHub Action for notifications
- RAG evaluation, vector search and retrieval quality
- ML training and inference pipelines with GPU acceleration
- Monitoring for Linux and GPU infrastructure
- CI/CD automation, security scanning and container workflows
Languages: Python, Go, PHP
Backend: Django, FastAPI, Flask, aiogram, maxapi
AI/ML: RAG, LangChain, Ollama, PyTorch, Keras, pandas, NumPy
Databases: PostgreSQL, MySQL, SQLAlchemy 2, Alembic, ChromaDB
Infrastructure: Docker, Docker Compose, Linux, GitHub Actions, Gitea Actions, Trivy
Observability: Prometheus, Grafana, TensorBoard
Desktop: PyQt6, Tkinter, Fyne
Tools: Git, Make, uv, ruff, pytest, pre-commit, Nginx, Caddy, Redis, Celery
Homelab: Ryzen 7 7700 · RTX 5070 Ti · 96 GB RAM · Ubuntu 25.10 — models, databases and monitoring all run here first, cloud second.
| Project | What it does | Highlights |
|---|---|---|
| ROoP | Fully local document assistant powered by RAG | Django, ChromaDB, Ollama, REST API, Docker, CPU and GPU support |
| Green Sentry | Plant disease diagnosis with computer vision and AI reasoning | FastAPI, ResNet50, GigaChat, React, PostgreSQL, Redis, Celery |
| Plant Disease ResNet50 | GPU-accelerated plant disease model training and inference | PyTorch, ResNet50, CUDA, Docker, TensorBoard, Jupyter |
| Stress Tester | Cross-platform CPU and RAM stability testing application | Go, Fyne, memory safety controls, Windows and Linux CI |
Bots and tooling for MAX — from a reusable project template to a GitHub Action that reports builds straight into a chat.
| Project | What it does | Stack |
|---|---|---|
| max-ollama | Chat with local Ollama models inside MAX; streamed replies, model switch, vision | Python, maxapi, Ollama, Docker |
| template-max-bot-github-gitea | Production-ready bot template with dual CI/CD for GitHub Actions and Gitea | Python, maxapi, uv, Docker, polling and webhook |
| max-action | GitHub Action that sends build, deploy and PR notifications to MAX | Go, zero dependencies, MAX Bot API |
| base-max-bot | FAQ bot with an in-bot admin panel, broadcasts and email notifications | Python, SQLAlchemy 2, asyncpg, Alembic, pytest |
| ID Helper Bot | Resolves user_id, chat_id and media IDs for MAX bot development |
Python, maxapi, Docker, uv |
| Project | Description | Stack |
|---|---|---|
| rocm-smi-exporter | Prometheus exporter and Grafana dashboards for AMD GPU metrics | Python, ROCm, Prometheus, Grafana |
| Tg-ollama | Telegram bot backed by local Ollama models | Python, aiogram, Ollama |
| monitor-dir | File integrity monitoring with Prometheus metrics | Python, Docker, Prometheus |
| arxiv2word | Converts articles from arXiv into editable Word documents | Python, Pandoc, Docker, Poetry |
| ga_detector | Chrome and Firefox extension for detecting analytics and verification tags | JavaScript, Chrome, Firefox |
| Fast-Chat-Auth | Chat application with registration and JWT authentication | FastAPI, JWT, Python |
| PHPNewsHub | News publishing and management system | PHP, PostgreSQL |
I'm interested in contributing to projects related to:
- self-hosted AI and local LLMs;
- RAG pipelines and LLM observability;
- Python and Go backend services;
- messenger bots and chat automation;
- monitoring and infrastructure tooling;
- Prometheus exporters and Grafana integrations;
- reproducible ML and MLOps workflows.
