← Back to projects

BillSage AI

Upload a bill — image or PDF — and get back a classified, structured output with confidence scores. Async, concurrent, fast.

PythonOCRFastAPIStreamlit
View on GitHub ↗

The problem it solves

Manual bill processing is tedious at scale. BillSage automates the classification and data extraction step — handling utility bills, grocery receipts, medical invoices, retail bills — and exposes the results through a Streamlit dashboard with accuracy metrics.

How it works

Image and PDF uploads are processed asynchronously through a concurrent pipeline. OCR extracts text from scanned documents. An LLM classifier categorises the bill type. The result — type, subtype, extracted data, confidence score — is surfaced in the dashboard. The system is async load-tested to confirm it holds up under concurrent requests.