Quote Automation & Smart Inventory Matching (under NDA)

Client: an e-commerce seller (ongoing engagement, under NDA).

Problem: every customer quote request arrived in an online inbox and had to be read, matched against stock and answered by hand.

What I built:

A Playwright-based reader that pulls new requests continuously and extracts the product details and condition from each request into structured data

A compatibility engine that scores each request against the seller's inventory on several product attributes and sorts matches into high, possible and insufficient confidence bands

Automatic offer drafts with pricing and product-specific messaging

A Next.js review dashboard where every offer is approved by a human before it's submitted, with no auto-sending

Local OCR that reads labels, barcodes and QR codes from customer photos

Production reliability: PostgreSQL job queue, parallel workers, retries with backoff, stale-job recovery and full audit trails

Outcome: a production system that turns a manual inbox into a reviewed, one-click quoting workflow, with more features on the roadmap.