AboutSkillsWorkProcessContact
Back to work
Retail · Auto PartsSaaS · Point of Sale2026

Motorepuestos Arias POS

A multi-branch point-of-sale platform engineered as a SaaS product for retail businesses of any size.

Motorepuestos Arias POS

Role

Sole Engineer — architecture, product & delivery

Industry

Retail · Auto Parts

Timeline

Client build, evolved into a reusable SaaS

Stack

12 technologies

Overview

What this project is

An auto-parts store owner needed software to replace a tangle of manual processes — tracking inventory, checking stock movements, invoicing, and keeping tabs on customers who buy on credit. Rather than building a throwaway one-off, I engineered the system as a SaaS product with a multi-branch, multi-warehouse, multi-register architecture so it can adapt from a single small shop to a larger multi-location business.

Context

Direct engagement with the owner of Motorepuestos Arias, a local auto-parts store, who needed to move day-to-day operations off spreadsheets and paper.

Main goal

Replace manual sales, inventory, and credit tracking with one reliable system — and make that system reusable for other retailers.

For business

The problem — and the value delivered

Written for owners and stakeholders: what was broken, why it mattered, and what changed.

The problem

Sales, inventory, invoicing, and customer credit were handled manually. There was no reliable view of stock on hand, no low-stock warnings, and credit balances lived on paper — which meant lost time, human error, and money that was easy to lose track of.

Why it mattered

For a parts retailer, stock accuracy and fast invoicing are the business. Every mistake is a missed sale, a wrong count, or a receivable that quietly disappears.

The solution

A web point-of-sale system that handles the full retail loop: fast invoicing, product management, real-time inventory with movement history, low-stock alerts, customer and supplier records, credit sales with a running receivables ledger, and role-based access for cashiers and admins. Built to run one store today and many tomorrow.

Benefits delivered

  • Real-time inventory with a full history of every stock movement.
  • Automatic low-stock alerts so nothing sells out unnoticed.
  • Fast, customizable invoicing tailored to the business.
  • Credit sales tracked per customer with a clear receivables balance.
  • Roles and permissions (cashier, admin) that keep operations safe.
  • One codebase that scales from a single shop to multi-branch operations.

Paper → Digital

Operations moved off spreadsheets

Real-time

Inventory & stock movements

Multi-branch

SaaS-ready architecture

For engineers

Architecture & implementation

Written for developers and recruiters: the technical decisions, tradeoffs, and lessons.

Architecture

The core design challenge was modeling a multi-branch, multi-warehouse, multi-register domain in a single, tenant-adaptable product. I approached it with SOLID principles and Clean Architecture, isolating business rules from the data and UI layers, and going through multiple refactoring passes until the code was polished and well-structured. Supabase provides Postgres, authentication, and object storage behind a typed data layer, with TanStack Query owning server state, caching, and invalidation on the client.

Frontend

  • React
  • TypeScript
  • Vite
  • Tailwind CSS
  • TanStack Query

Backend & Data

  • Supabase
  • PostgreSQL
  • Supabase Auth
  • Supabase Storage (S3)

Testing

  • Vitest
  • Playwright

Delivery

  • Vercel
  • Git & GitHub

Implementation highlights

Multi-branch / multi-warehouse / multi-register model

A single data model that cleanly separates branches, warehouses, and registers so stock and sales stay correct across locations — the hardest and most valuable part of the build.

Role-based access control

Cashier and admin roles with scoped permissions, enforced through Supabase Auth and row-level policies.

Inventory movement ledger

Every stock change is recorded, giving an auditable history and driving real-time low-stock alerts.

AI-assisted engineering

I used AI assistants (Claude with Opus models, and Gemini) to pressure-test design decisions — comparing tradeoffs, hunting for weak points, and reviewing refactors — while keeping full ownership of the architecture.

Challenges

  • Keeping stock correct across many locations

    Multi-warehouse inventory means concurrent movements can corrupt counts. I modeled stock as an append-only movement ledger so quantities are always derivable and auditable.

  • One product, many businesses

    Turning a client tool into a SaaS meant designing for configurability without special-casing. Clean Architecture boundaries let the same rules serve a tiny shop or a multi-branch operation.

Lessons learned

  • Modeling the domain first — before the UI — pays off across every later feature.
  • An append-only ledger beats mutable counters for anything money- or stock-related.
  • AI review is a force multiplier when you stay the decision-maker, not a passenger.
Features

What it does

Fast Invoicing

Customizable invoices built for quick checkout.

Inventory Control

Real-time stock with full movement history.

Low-Stock Alerts

Automatic warnings before items run out.

Multi-Warehouse

Stock tracked across branches and warehouses.

Roles & Permissions

Scoped access for cashiers and admins.

Credit & Receivables

Per-customer credit sales and balances.

Technologies

Full technology stack

ReactTypeScriptSupabasePostgreSQLSupabase AuthSupabase StorageTanStack QueryViteTailwind CSSVitestPlaywrightVercel

Want something like this for your business?

I turn manual, error-prone workflows into reliable software. Let's talk about what you need.

Start a conversation