Anton Chuprun

Anton Chuprun

Python Backend & AI Automation Engineer. Open to part-time and full-time contracts.

years of shipping
6
released products
10+
open-source repos
6
end-to-end delivery
100%

Selected work

Zynerio
Zynerio
VPN service run solo: bot, backend, web & fleet automation
View
Zynerio VPN landing (zynerio.com)
Web cabinet: active subscription & 1-click connect
In-bot admin panel: grants, promos, refunds, analytics
Tariff picker in the Telegram bot
SBP QR payment flow in the Telegram bot

VLESS/XRay VPN sold through a Telegram bot (aiogram): plans, promos, referrals, refunds, Platega QR payments, in-bot admin with analytics. Web cabinet on React + FastAPI (JWT, Telegram Login). Remnawave is the only third-party piece (panel, config distribution, user DB); nodes run remnanode + Cloudflare WARP. The fleet is managed by a custom CLI over Ansible: one-command node bootstrap, rolling updates, DNAT front-relays, backups, healthchecks.

Live product PythonFastAPIaiogram 3ReactPostgreSQLAnsibleDockerVLESS / XRay
VC
Video Conveyor
Autonomous short-video farm: any source → viral clips for TikTok / Reels / Shorts
View
Ready clip: hook title, karaoke subs & animated banner
Ready clip from another source: subs & banner overlay
Ready clip: per-platform render for TikTok / Reels / Shorts

Short-form video farm promoting Zynerio VPN. Any source (YouTube channel/video or local files) is fetched by yt-dlp with bot-wall bypass and transcribed by faster-whisper large-v3 on GPU with word-level timings. A local LLM (Ollama) picks viral moments, verifies story boundaries and refines cuts by waveform. FFmpeg renders 9:16 blur-pad clips with word-by-word pop-in subtitles (libass), a censorship filter, programmatically rendered animated banners, and per-platform uniqueness packs (background, speed, crop jitter, mirror) for TikTok / Reels / Shorts. Jobs run as an on-disk state machine with a FIFO scheduler and a ledger deduping already-cut scenes.

Automation PythonFastAPIWhisperLLM / OllamaFFmpegyt-dlpDocker
AI Chan
AI Chan
AI anime companion with Live2D, emotions & voice
View
AI Chan object recognition & pricing
AI Chan calorie counting with emotional reaction
AI Chan gaming slang dialogue & emotions
AI Chan translator mode & TTS speech

Interactive chat companion with animated Live2D characters: real-time lip-sync, TTS/STT voice dialogue, vision, and image generation. Full mobile production stack: in-app subscriptions, ads, analytics, push notifications, and WebSocket streaming.

Live product FlutteriOSLive2DTTS / STTIAP / SubscriptionsAdMobAnalytics
Cozy Coloring
Cozy Coloring
Drawing & coloring app with custom canvas and pixel-perfect palettes
View
Cozy Coloring rabbit coloring canvas & palette
Cozy Coloring watermelon brush & opacity controls
Cozy Coloring cozy cafe scene drawing workflow
Cozy Coloring landscape coloring & precision tools

Drawing app built with Flutter and a high-performance custom canvas engine. Features intuitive brush controls, opacity layering, dynamic color palettes, and seamless in-app purchases via RevenueCat & StoreKit.

App StoreFlutter FlutteriOSCustom CanvasStoreKitRevenueCat

Open source

All repositories ↗

About & Experience

Background

I design and ship production-ready async backends with FastAPI, Telegram bots, and distributed data processing systems. I handle features end-to-end: from PostgreSQL database schemas, indexing, and Redis caching to CI/CD pipelines, Docker containerization, and AI model orchestration.

Experienced in building GPU media pipelines (Whisper large-v3, FFmpeg NVENC, local LLMs) and automated server fleets. Async-first by default, comfortable owning services without micromanagement, and focused on shipping working software to production.

years of shipping
6
released products
10+
open-source repos
6
end-to-end delivery
100%
Remote · Part-time (priority) & Full-time Async-first, text-based workflow UTC+2 (Kaliningrad) USDT / USDC / Bybit UID / Mellow / 4dev

What I do

FastAPI, Backend & Microservices

Async REST APIs and WebSockets, transactional integrity with row-level locks in PostgreSQL to prevent double-spending, Redis caching, Alembic migrations, third-party API integrations, and billing webhooks.

AI, Media Pipelines & Automation

GPU-accelerated video/audio processing pipelines (Whisper large-v3, FFmpeg NVENC, dynamic ASS subtitles), Telegram bots with analytics (aiogram 3), async data scrapers, and automated server fleets via Ansible.

Start with a fast paid pilot

2–3 days with a fixed price and clear scope to test team chemistry, build a high-impact feature or prototype, and seamlessly transition into a retainer or full contract.

Get in touch

Send a direct message

Drop a note directly to my Telegram inbox. Fast response guaranteed.

I receive messages instantly in Telegram