← Back to projects
Full Stack Developer

Baby Pro - Textile Manufacturing & Inventory System

PT Hero Busana Textindo · Team — 3 people · 2025 - Present

  • React.js
  • Python
  • FastAPI
  • PostgreSQL
  • Tailwind CSS
Internal project (not public)
1 / 6

My contributions

  • Dominantly handled frontend development, including UI implementation, frontend logic, and FE-BE integration
  • Built core backend functionality while CRUD implementations were also handled by a junior developer
  • Translated complex manufacturing business flows into system workflows and production states
  • Reviewed and corrected junior developer implementations when business-flow requirements were misunderstood
  • Built production planning and tracking flows for cutting codes across printing, embroidery, and pants production
  • Implemented monitoring flows for CMT deliveries and outstanding production batches
  • Contributed to the Warning Dashboard for monitoring payment and delivery conditions
  • Implemented server-side idempotency middleware to prevent duplicate write operations from network retries and double-clicks
  • Optimized product listing image loading by generating WebP thumbnails at upload time, reducing per-page image payload from 10–40MB to under 300KB

Overview

Built an internal textile manufacturing and inventory system for tracking materials and products from fabric intake through cutting, production processes, and finished goods. The system supports production planning and traceability across printing, embroidery, and pants production workflows, including partial processing and CMT delivery monitoring.

✦

Multi-stage textile production tracking

✦

Fabric-to-finished-goods inventory flow

✦

Production and CMT delivery monitoring

Challenges

  • Tracking products across multiple production stages and CMT suppliers
  • Handling partial production where different cutting codes could be at different stages
  • Replacing manual cutting and CMT tracking that made follow-up difficult
  • Ensuring production codes could only move through valid workflow stages
  • Preventing duplicate data mutations from network retries and accidental double-click submissions on write operations
  • Product listing pages loading full-resolution images (10–40MB per page) caused slow renders and frequent aborted requests

Solutions

  • Mapped production stages into explicit planning and status flows
  • Tracked cutting codes and quantities throughout the production process
  • Added CMT delivery monitoring to identify outstanding production batches
  • Built Warning Dashboard views for payment and delivery follow-up
  • Built a global FastAPI middleware that intercepts POST/PUT requests, checks a PostgreSQL-backed cache by idempotency key, and returns the cached response on duplicate requests — transparent to all route handlers
  • Generated WebP thumbnails at upload time using Pillow and served them via a dedicated route; frontend resolves thumbnail URLs from original paths and falls back gracefully to original or placeholder icon

Outcomes

  • Improved visibility across the textile production workflow
  • Made cutting and CMT progress easier to track
  • Provided management with a centralized view of payment and delivery warnings
  • Eliminated duplicate submissions — repeated write requests return identical cached responses without re-executing business logic
  • Reduced image payload per page load by ~99%, from 10–40MB down to 100–300KB, eliminating abort-prone requests on the product listing page