FOBOH Pricing Profile Challenge

A full-stack TypeScript application for managing wine/beverage pricing profiles. Users configure pricing rules, select products, assign customers, and review before submitting.

Backend

Stack: Express 5, TypeScript, Zod, Fuse.js Hosting: Railway

Architecture

The backend follows clean/layered architecture with three layers:

• Domain (domain/types.ts) — Entity types (Product, Customer, PricingProfile, etc.) and Zod schemas for request validation. No framework dependencies.
• Infrastructure (infrastructure/database.ts) — Data access layer built as an adapter over a JSON file. Reads db.json into memory on startup and persists mutations back to disk. Exposes only typed read/write functions, so swapping to a real database would only require replacing this file.
• Interfaces (interfaces/*.router.ts) — Express routers, one per resource. Thin layer that calls into infrastructure and returns responses.

Database

Instead of a real database, data lives in a JSON file (db.json). An in-memory adapter sits on top of it, providing:

• Typed read functions with denormalization (e.g., products are returned with brand names resolved from IDs)
• Fuzzy search via Fuse.js (threshold 0.4, searching across title, SKU, category, segment, brand)
• Write functions that validate with Zod, then persist the updated state back to the JSON file

Security

Basic security measures via middleware:

• Helmet for security headers
• CORS with an explicit origin allowlist (configured via ALLOWED_ORIGINS env var)
• Rate limiting (100 requests per 60 seconds)
• Request body size limit (1MB)
• Zod validation on all POST request bodies

Frontend

Stack: Next.js, React, TypeScript, Tailwind CSS, shadcn, React Query, React Hook Form Hosting: Vercel

Structure

Single page application matching the spec. The main component (pricing-setup.tsx) is a multi-step wizard:

1. Basic pricing profile configuration
2. Product selection with search, filters, and price adjustment preview
3. Customer assignment
4. Review and submit

• UI: Built with shadcn components.
• Data Fetching: React Query manages server state. Search inputs are debounced (300ms) before triggering API calls.
• Form state & Validation: Managed by React Hook Form with Zod schema validation.
• Price Calculations: Performed server-side via POST /products/calculate-prices. The frontend sends product IDs, adjustment mode/direction, and per-product adjustment values; the backend resolves base prices, applies the formula, and returns calculated results. The "Refresh New Price Table" button triggers this call.

Expansion Considerations

The frontend communicates directly with the backend. In a larger application, I'd introduce a BFF (Backend for Frontend) layer between them for Nextjs. A BFF handles response transformation, aggregation across services, and keeps the frontend decoupled from backend domain models.

Use of AI

Claude was used as a development aid during this project for scaffolding boilerplate code, writing tests, and accelerating implementation tasks. All architecture decisions, code review, and final implementation were my own.