open via a local server to see 3D decor
pick a floppy · or press 1–5
AI Product Engineer — LLM agents, evals & applied AI
I build AI agents that talk to real customers.
Then I build the evals that keep them from embarrassing anyone.
Full-stack is the substrate, not the pitch. I'll own the model layer, the backend, the UI and the 3am pager — but the job is making LLMs reliable at scale.
● Open to AI engineering roles · remote
< Download CV >C:\> █
The chat widget where an AI agent sells for e-commerce brands. Mine since 2024.
Result: the AI chat real shoppers use, every day, on real stores.
AI that answers a brand's comments and DMs on social — in the brand's voice, with the right product attached.
Why it matters: comments get a real answer while the social team sleeps.
An action layer for AI agents: ask for a dinner spot, a gym or a spa — the agent finds it and books it.
Measured: tasks done 37% faster (5:37 vs 8:54). Projected: 42–62% lower API cost.
AI lead generation for WhatsDo: agents that find businesses worth bringing onto the platform.
A sales team of one. Plus agents.
Multimodal AI on Web + Telegram: chat, images, voice, memory. Built and run solo — model layer to billing.
~3K people a day · 99.9% uptime · first token in 50–150 ms · infra bill cut 3×
Open models on my own GPU — not just someone else's API key.
Also heats the room in winter.
A pipeline that turns gameplay footage into finished vertical shorts — voiceover, captions, cuts — for faceless channels.
Result: a ~30-second short from a script, without opening a video editor.
A real-time trading bot that reads Binance and OKX order books and hunts price gaps between them.
Runs on paper money only. I like my savings.
A game built by 8 coding agents working in parallel, with reviewer agents checking each other's work.
I was the engineering manager. Nobody asked for a 1:1.
Built and shipped a game in Unity, start to finish.
Proof I finish side projects. Sometimes.
arteom@workstation:~$ git log --career --no-pager
arteom@workstation:~$ █
Hiring for AI engineering — agents, evals, LLM products? I reply faster than this modem.