← back to portfolio
case study· 2024· Product Engineer· Full-stack build

InvoiceIQ

AI-powered invoice management — parse, structure, sync.

📄
Overview

InvoiceIQ ingests messy invoice PDFs and images and outputs three synchronized views — Invoices, Products, Customers — so a single update propagates everywhere.

Challenge

Finance/ops teams spend hours re-typing invoices and reconciling mismatched product and customer records across tools.

Outcome

Structured, synchronized data from unstructured invoices — the manual-entry step disappears.

Approach

How it was built.

  1. 01

    Built a document-parsing pipeline combining Google Generative AI for structured extraction, PDFPlumber for text PDFs, and Tesseract OCR for scanned images.

  2. 02

    Modeled invoice/product/customer as a single graph in Redux Toolkit, so an edit in one view updates the others.

  3. 03

    Shipped a React + Vite + TypeScript frontend against a FastAPI backend for durable extraction jobs.

Stack
ReactViteTypeScriptRedux ToolkitFastAPIGoogle Generative AIPDFPlumberTesseract OCR
GitHub →← All projects