Vlad Korzinin

AI Builder & Product Engineer

I take an idea to a live product — the design, the code, and everything underneath.

For 15 years I've built things end to end — a physical creative brand and all its software, and this past year, AI products: a live research-memory tool with an MCP server, an AI tutor, internal tools, and agent experiments. I design it, build it, and run it.

Currently building Stillscase in Barcelona. Open to AI product, enablement, and automation roles — remote or relocation. Get in touch → e-maillinkedin


Selected work

Stillscase

A research-memory tool for people who work inside AI assistants. A remote MCP server lets Claude, ChatGPT and Cursor search and cite your own sources — powered by an ingestion pipeline (YouTube, audio, PDFs), transcription, and hybrid full-text + embedding search, with multi-tenant isolation. Live, with free and paid tiers, running in production.

  • MCP server
  • RAG
  • Embeddings
  • Next.js
  • Chrome extension
  • OAuth
View project
Mia

MIA

An AI co-pilot for independent language tutors. A Plan → Deliver → Capture → Improve loop: analyze a lesson recording, update long-term student memory, and generate the next lesson plan from fresh evidence. Multi-stage AI orchestration with plan judging and regeneration, plus a prediction-and-grading loop that scores plan quality, Stripe billing, and an experimental voice-practice mode.

  • LLM orchestration
  • Transcription
  • Longitudinal memory
  • Next.js
  • Supabase
View project

Ships

An internal operations tool that ran fulfillment for my own commerce business. Syncs Shopify orders and automates shipping, tracking, PDF invoicing and accounting — so a two-person team could ship products worldwide. Includes browser automation for a shipping provider with no clean API.

  • Ops automation
  • Shopify
  • Playwright
  • PDF generation
  • Tauri
Private

Vibespace

An experiment in agent orchestration. A multi-agent pipeline that turns a Slack request into reviewed, merged code — agents triage the request, write a spec, implement it, review the pull request, and either merge or escalate, with a durable event bus, retries, and self-healing. A working system I built to see how far this can go, not a finished product.

  • Multi-agent
  • LLM orchestration
  • Slack
  • GitHub
  • SQLite
  • Self-healing
Experiment

Playing Arts

A creative brand and e-commerce business I founded and still run — fully automated now, after a decade of building it end to end. I owned art direction, community, product, and technology myself: I launched illustrated-deck art contests, grew a community of 1,011 artists, and crowdfunded production across four funded Kickstarter campaigns (3,418 backers, ~$100k raised) with no marketing budget. I ran the whole operation — physical manufacturing, global logistics and fulfillment, marketing (Instagram and email), a 17-episode podcast, the full website design and development, and an augmented-reality app that animates the printed cards when scanned. A later digital collectible drop did $17.4M in secondary volume and hit #1 on OpenSea on launch day, ahead of CryptoPunks and Bored Ape Yacht Club.

  • Founder
  • End-to-end ownership
  • Physical production
  • Logistics
  • Marketing
  • Web design & dev
  • AR app
  • Crowdfunding
  • Community
  • NFT
View project

Crazy Aces

A card game built around the Playing Arts deck — web and mobile (React Native), shared game logic, a server-authoritative engine, and growth mechanics: performance-based discount rewards, email capture, and conversion analytics with anti-abuse controls.

  • React Native
  • Game engine
  • Serverless
  • Redis
  • Growth mechanics
  • iOS
View project