← Back to work

Restaurant SaaS • Founder

VIVORA

Orders without waiters, apps or a POS terminal.

VIVORA cover
Role
Founder, sole engineer
Company
Independent
Period
2026 — NOW

Context

Built for Uzbek-run restaurants in South Korea, where guests and cooks don’t share a language.

Guests scan the QR code on their table, browse a live menu in their own language, and order from their phone. The ticket reaches the kitchen display in under a second and prints on the thermal printer by the pass. Owners run menus, tables, running tabs and revenue from one dashboard, and the platform invoices each restaurant a flat monthly fee.

I run it end to end: infrastructure, releases, customer onboarding and support.

At a glance

Restaurants live
8
Languages
4
Order to kitchen
<1s
Apps to install
0

Key features

  • QR ordering, zero install

    Scan a table code, order in your language, watch the status change live. No app, no account.

  • Real-time kitchen display

    Orders arrive on an always-on board over WebSocket subscriptions and print automatically.

  • Running table tabs

    The first scan opens a tab and every later order joins it; a database-level index guarantees one visit never splits into two bills.

  • One kitchen language

    Guests read English, Korean or Russian. Tickets and reports stay in Uzbek, because that’s what the cooks read.

  • Self-driving lifecycle

    Orders accept, serve and settle themselves on a timer, so service completes even when nobody touches a screen.

  • Owner analytics

    Revenue, peak hours, best sellers and average ticket, computed in MongoDB aggregation pipelines.

Architecture

A cloud GraphQL API, and a local print agent that bridges it to the kitchen.

A NestJS + TypeScript GraphQL API (code-first) with MongoDB, JWT auth and role guards pushes live order feeds over graphql-ws. The Next.js 14 + MUI frontend uses Apollo Client and Zustand.

A standalone Node print agent, installed as a Windows service, drives ESC/POS thermal printers on the restaurant’s own LAN. Menus are translated with the Gemini API, staff get Telegram alerts, and the stack deploys on Railway and Vercel.

Panels

A restaurant dashboard for owners, and a super-admin panel for the platform.

  • Orders & tables

    Live orders, phone orders, payments, table setup, QR codes and closing bills.

  • Menu & availability

    Categories and dishes in four languages, prepped portions that sell out automatically at zero.

  • Billing & feedback

    Monthly invoices, bank-transfer reporting, and guest ratings and comments.

  • Approval-gated signups

    Restaurants stay locked until a super admin approves them, so nobody goes live unreviewed.

  • Suspend without rejecting

    An approved restaurant can be switched off and back on without losing its approval.

Used technologies

TypeScript · NestJS 10 · GraphQL + graphql-ws · MongoDB 7 · Next.js 14 · MUI v5 · Apollo Client · Zustand · ESC/POS print agent · Gemini API · Telegram Bot API · Docker

Scope & responsibilities

Product · Full-stack build · Print agent · Operations