<h1>

Hey
I'm Denis,
Fullstack Developer

</h1>

<p>

I build production-ready systems solo — from idea to deployed product in 1–3 weeks. Python backend, Next.js frontend, Docker DevOps, AI/ML infrastructure. 5 systems shipped. Open to remote roles globally.

</p>

Let's Talk

Production
Projects

Years
in IT

English &
Bulgarian

About Me

Non-linear path to code

<p>

Hello!
My name is Denis and I specialise in fullstack development using Python, Django, FastAPI, Next.js, and AI/ML infrastructure.

I'm a highly motivated individual dedicated to writing clean, concise, robust code that works. Linguistics graduate from Moscow State University who became a full-time developer — striving to never stop learning.

My path: MSU linguistics → software localisation at OpenText & Hummingbird → systems admin in Bulgaria → industrial robotics → and since 2024, full-time Python & AI/ML.

I run OWASP Top 10 security audits on my own code, deploy with Docker Compose + Nginx + Let's Encrypt SSL on every project, and write structured logs and health checks from day one.

</p>

PythonDjango FastAPIFlask Next.jsReact TypeScriptDocker NginxPostgreSQL RedisAI/ML LLMFlutter

Skills

I am striving to never stop learning and improving

Web Development

Django · FastAPI · Flask
Next.js · React · TypeScript

AI / ML Infrastructure

LLM deployment · GGUF
llama.cpp · Ollama · CUDA

DevOps & Infra

Docker · Nginx · Linux
SSL/TLS · PostgreSQL · Redis

Telegram Bots

aiogram · FastAPI
YooKassa · WebhooksAsyncSQLAlchemy

Python

Django

FastAPI

Next.js

React

TypeScript

🐳

Docker

Nginx

PostgreSQL

Redis

Projects

5 production systems shipped solo — idea to deploy in 1–3 weeks

NoSpoon Messenger

View →

Anonymous cross-platform messenger with Brainfuck encoding, IPFS storage, Cloudflare Workers relay, CRT terminal UI. Flutter + Dart.

Flutter 3.27 Dart Brainfuck compiler IPFS (Pinata v2)

SochiSpirit Marketplace

View →

Handmade goods marketplace with WebSocket chat, 5 user roles, moderation, Telegram Login. Built in 2 weeks.

Django 5.1 DRF Django Channels Celery

View All Projects →

Saiga LLM

Production-ready multi-user LLM service on a self-hosted GPU cluster

Saiga Nemo 12B — on local hardware

Saiga Nemo 12B (Llama family, GGUF Q4_K_M) running locally on 2× NVIDIA P104-100 with layer offloading via llama.cpp. ~7.5 GB total VRAM distributed across cards, 15-18 tokens/sec at 8K context. OpenAI-compatible API, web UI and Telegram bot — distributed deployment with full observability stack (Prometheus + Grafana + DCGM).

saiga.vaibkod.ru @saiga_ai_bot github.com/saiga

12B

Parameters

Context window

15-18

Tokens/sec

~7.5GB

Total VRAM

Learn more about Saiga →

GPU Cluster

2× P104-100 mining GPUs with layer-split via llama.cpp. CUDA 11.8 in container. Third P104 in the same host repurposed for Stable Diffusion in a sibling RAG project.

Distributed Deployment

Bot on separate VDS via SSH-tunnel + restricted-key. LLM via HTTPS+Bearer.

Production Observability

Prometheus + Grafana + DCGM-exporter. GPU memory/util/temp/power dashboards.

CI/CD + Tests

GitHub Actions with ruff + pytest. 62 unit tests, 99% coverage on tested modules.