Car Dealership Management System
A full-stack platform that runs a car dealership end to end - inventory, quotations, sales, installment financing, collections, disputes, supplier invoices and financial reporting - built for a market where cars are sold on long installment plans and every figure has to hold up to an audit.
Client
unknown
Type
System
Timeline
unknown
Deliverables
Express and TypeScript REST API React and TypeScript admin interface PostgreSQL schema with versioned migrations Role-based user and permissions management Installment financing and collections workflow Financial reporting with Excel and PDF export Sorani Kurdish localization layer Automated regression and financial verification suites Setup documentation and client go-live checklist
THE CHALLENGE
The dealership's operations lived across disconnected spreadsheets. Every car carried a different cost basis — purchase price, Dubai import costs, service work, linked expenses — so nobody could say what a single sale actually earned. Installments were tracked by hand, late fees were applied inconsistently, and monthly totals never reconciled. The hard requirement was that money could only have one source of truth: a figure recorded at the moment of sale is historical truth and must never silently change when a formula is corrected later.
THE SOLUTION
An Express and TypeScript API on PostgreSQL with Knex migrations, JWT access and refresh tokens, role-based access for admin, manager and staff, and scheduled jobs that process overdue installments and recalculate client risk overnight. The interface is a React 18 and Vite single-page app using TanStack Query, Zustand, React Hook Form with Zod validation, Tailwind and Recharts. Every business-financial formula lives in one authority module, mirrored on the frontend for previews only, and an automated check fails the build if a calculation leaks anywhere else. On top of that sit multi-step quotation and sale wizards, automatic installment schedule generation, day-based late-fee calculation, per-car profit analysis, Excel and PDF financial exports, and a Sorani Kurdish runtime translation layer covering the whole interface. The regression suite creates a temporary PostgreSQL database, runs the real migrations against real SQL and real service logic, then drops itself - so the financial tests prove the actual queries, not mocks.
THE RESULT
One system now answers the questions the spreadsheets could not: what a specific car really cost and earned, which clients are risky based on late payments and open disputes, and how a month's revenue stands against its expenses in a way that reconciles with the P&L. Late fees apply identically every time, and inventory rolls back correctly when a sale is cancelled. Roughly 26,000 lines of TypeScript across 109 files and 26 screens. Handover is gated: a financial verification script must return zero issues, the regression and contract suites must pass, and the demo seed must be provably absent before the system goes live.