Back to Projects

Unified POS

AI-Assisted Point of Sale System

Jan 2026 – Mar 2026

Professional Project – Developed at FrontendEx

Unified POS screenshot
Unified POS screenshot
Unified POS screenshot
Unified POS screenshot
Unified POS screenshot

Overview

A point-of-sale platform built for retail stores that need to keep selling even when the internet or the central server drops — a local-first desktop client that syncs across registers on the store's own network and reconciles with the cloud once connectivity returns, alongside a web admin dashboard, a mobile companion app, and AI-assisted inventory insights.

  • Developed a centralized point-of-sale platform handling sales processing, inventory tracking, and vendor management, using Redis caching to speed up frequent product lookups.
  • Integrated the Llama API to generate stock insights and vendor recommendations, cutting manual procurement effort through automated suggestions and real-time analytics.

Approach

  • Architected as a local-first desktop client that keeps selling, printing receipts, and tracking stock entirely offline, then reconciles with a central API once connectivity is restored.
  • Added a LAN sync host so multiple registers on the same store network stay in sync with each other in real time even when the internet is down, not just eventually-consistent with the cloud.
  • Used Redis to cache frequent product lookups on the hot checkout path, and integrated the Llama API to turn raw sales and inventory data into stock and vendor recommendations for store owners.
  • Structured the codebase as a Turborepo monorepo (desktop client, web admin dashboard, mobile companion app, API, sync host) sharing types and business logic across surfaces instead of duplicating them per app.

Technologies Used

Next.jsNest.jsSupabaseRedisLlama API