class MLEngineer:
def __init__(self):
self.name = "Denis Kustov"
self.role = "ML Engineer · Data Scientist · Annotation Lead"
self.location = "Saint Petersburg, Russia 🇷🇺"
self.focus = ["LLMs", "RAG Pipelines", "Healthcare AI", "Data Annotation"]
self.stack = ["Python", "PyTorch", "Docker", "PostgreSQL", "Proxmox"]
self.research = "Annotation Pipeline Agents (Master's thesis @ ITMO)"
self.currently = "Building intelligent data pipelines & LLM agents"
def fun_fact(self) -> str:
return "I run my own home lab on Proxmox 🖥️ and drink too much coffee ☕"What I'm actively working on — updated May 2026
- 🎓 Writing Master's thesis on automated annotation pipeline agents (ITMO)
- 🤖 Designing multi-agent RAG systems for medical document processing
- 📊 Exploring LLM-as-judge evaluation frameworks for annotation quality
- 🏠 Expanding home lab: Proxmox cluster + self-hosted AI inference stack
- 📖 Preparing for English C1 interviews in international ML teams
| Project | Description | Stack |
|---|---|---|
| 🤖 rvenie_bot | Mistral-powered GPT Telegram bot | Python · Mistral AI · Telegram API |
| 📡 nsi-client | Client library for NSI data API | Python · REST |
| 🎵 listenurl_bot | Music link converter for Telegram | Python · Telegram Bot API |
Languages & Core
ML / AI / LLM
Data & Annotation
Infra & DevOps
"The goal of ML is not to build the most complex model, but the most useful one."