Automated multi-vendor inventory → normalized catalog → ops-ready exports
A full-stack automation platform that scrapes vendor portals, normalizes units, and continuously updates inventory for a medical supply business.
Problem
Inventory lived across multiple vendor portals and spreadsheets. Updates were manual, error-prone, and too slow to keep pricing and availability accurate.
Stack
- - Python
- - Django
- - PostgreSQL
- - Playwright
- - Selenium
- - Workers/Cron
- - OneDrive API
- - OpenAI API (fallback parsing)
Architecture
Pipeline architecture
A simplified view of how the system is wired. (Details and metrics are intentionally omitted until confirmed.)
- →
Step 01
Vendor portals
Browser automation + login
- →
Step 02
Extract + normalize
Parsing, UoM normalization
- →
Step 03
Store + validate
Idempotent upserts in Postgres
- →
Step 04
Exports
Excel/OneDrive delivery
- →
Step 05
Ops UI
Admin dashboards + logs
- - TODO: Add real throughput + accuracy metrics once measured.
- - TODO: Add an Upwork review link if available for this engagement.
Outcome
What changed
- Centralized inventory updates behind a single UI and pipeline
- Repeatable scheduled runs with logs and failure visibility
- Safer normalization with deterministic rules + LLM fallback
Metrics
Metrics are shown only when verified. Placeholders live in `lib/data.ts`.
Media
Demo / screenshots
Walkthrough demo (click to load).
Want this kind of system for your business?
Share what you're building and we'll respond with a concrete plan and next steps—no generic proposal templates.