Portfolio

A full-stack production costing system built and deployed for a manufacturing client. The system replaces a manual 20–30 minute quoting process with a sub-10-second automated calculation — covering material costs, purchasing optimisation, pricing tiers, and profit margins.

The client needed a reliable way to cost any production run without specialist knowledge or complex spreadsheets. The system handles all the business logic behind the scenes and presents a clean, actionable summary.

Technical highlights

• IQR outlier detection — statistically groups 300+ products into fair pricing tiers, isolating anomalous pricing that would skew averages
• Dynamic programming purchase optimiser — finds the cheapest combination of available stock sizes for any required quantity. Replaced a recursive brute-force approach that struggled at scale; DP version handles 10,000+ unit runs in milliseconds
• Tier pricing formula — MAX(average, highest × 1.10) guarantees no losses while keeping prices competitive across each tier
• Flask REST API — clean separation between backend logic and frontend, three endpoints serving the full application
• CSV → JSON pipeline — supplier pricing CSVs are ingested, reshaped, and stored as structured JSON at startup. Updating prices requires only editing a spreadsheet

Impact

• Quoting time: 20–30 minutes → under 10 seconds
• Consistent tier-based pricing across 300+ products
• Full visibility into actual vs tier cost and real profit per job
• Purchasing optimisation prevents overstocking