EN · RU
AI infrastructure · web platforms · realtime & 3D · 30+ launches
I'm a product engineer. I take an idea and carry it to "it works for real users": I design the architecture, write the backend and the frontend, wire up model inference, payments and monitoring, then keep it alive in production.
30+ shipped products in two years — an AI model marketplace billed per token, an autonomous Windows agent, a 3D browser game, an AR building tool, a retail banking platform, client portals and brand sites. Some source code is closed by NDA or commercial confidentiality: there I show the product and explain the engineering decisions, not the code itself.
Principle: a working user journey first, beautiful architecture second. A product without users is just a folder of files.
A marketplace where authors upload model weights and inference runs on first-party GPUs: a catalog with per-million-token pricing, a web chat billed by actual usage, an OpenAI-compatible API and one invoice across every model. Billing, inference orchestration, response streaming, multi-tenancy — all built from scratch.
Role: architecture, backend, frontend, infrastructure Status: live, closed source — demo on request
An agent that operates a computer: plans tasks, works with apps and files, keeps context between sessions and runs autonomously for up to 10 days. I built the whole product — positioning, landing page, design system, copy, interactive demo and the Windows build.
Stack: TypeScript, React, Vite, hand-written CSS/Canvas animation Hard part: explaining "an agent on your computer" without hand-waving — through scenarios, not feature lists
Web-swinging between skyscrapers: custom rope physics, a procedural city, dynamic lighting and streaming of models and textures. The asset pipeline is a point of pride too — a PBR character with normals and metalness that doesn't wreck performance.
Stack: JavaScript, Three.js/WebGL, GLSL shaders, custom asset loader Hard part: holding 60 FPS on weak laptops — polygon budget, batching, lazy sector loading
AlphaBot Quantum — control room for a trading agent: capital, analysis cycles, live event stream, log export.React · TypeScript · Firebase Live · Code |
VILIS — launch site for premium AI glasses: 1080p capture, 8 hours of battery, built-in assistant.Brand site · conversion · motion Live |
AR 3D Builder — hand-tracked AR editor: pinch to place blocks, gesture to delete, scene export to JSON.WebXR · MediaPipe · Three.js Live |
ABEBA Bank — zero-fee meme bank: a product experiment in brand voice, virality and conversion.React · landing · Web3 Live · Code |
Easgik — customer portal for a land-survey and cadastre service: requests, documents, account area.Commercial client · Next.js · Postgres Live |
Barber Boss — premium barbershop in Astana: services, interior, booking flow, Telegram notifications.Client site · booking · brand Live |
GameViki — video game encyclopedia: catalog, reviews, ratings, collections, editorial admin.React · Node · Postgres · Redis Closed source · private beta |
BankFlow — retail digital banking: accounts, payments, cards, lending, investments, analytics.FastAPI · Next.js · Kubernetes Closed source · commercial |
| Product | What it is | Link |
|---|---|---|
| AInvest Terminal | Investment terminal: portfolio, daily yield, auto-invest 10% into stocks, insights feed | Live |
| SmartLearnix | AI learning platform: personal tracks, gamification, analytics | Code |
| TgClone | Messenger client on React: chats, media, realtime updates | Code |
| freeGPT | Zero-setup interface to language models | Code |
| AirOS | Browser "operating system": windows, files, apps in a tab | Code |
| CryptoMain | Crypto portfolio tracker with auth and live rates | Live |
| Stranger Things Shop | Fandom merch store: catalog, cart, checkout | Live |
| Alpha Barbershop | Second barbershop brand site with online booking | Live |
Six products run with private repositories — NeuroHub, BankFlow, GameViki, AInvest, plus two client systems under NDA. That's a normal part of commercial work: the product is public, the code is the client's asset.
If you need to see engineering depth rather than a landing page, I'm happy to walk through architecture, data models and trade-offs on a call — or show the open-source ones above, which are real and complete.
I also handle everything around the code: product decisions, interface design, copy, analytics, deployment and production support.
| Step | What happens |
|---|---|
| 01 | I turn the task into a user journey: who, why, what the outcome must be |
| 02 | I build the skeleton: data model, API, key screens — validated on real data, not mockups |
| 03 | I ship it: tests on critical paths, logs, metrics, error handling, deploy |
| 04 | I stay after release: iterations, handover of code and documentation |















