AI Invoice Processing & Data Extraction
Budget / SalaryA$1,500–3,000
TypeFreelance project
LocationRemote
Posted1 hour ago
I need an end-to-end AI solution that can ingest incoming invoices, pull out key information with at least 95 % accuracy, rename each file intelligently, and store everything in an organised folder structure (or another storage layer you recommend).
Input formats are varied—PDFs, JPEG/PNG photos taken on phones, and scanned documents—so the pipeline has to be format-agnostic. My initial thought is a machine-learning-driven extractor that combines OCR with layout analysis, but I’m happy to hear alternative strategies if they boost reliability or speed.
Here is the data I must capture on every invoice:
• Supplier name
• ABN
• Invoice number
• PO number
• Invoice date
• Total amount and GST
• Line-item stock details (quantity, description, unit price or total value)
After extraction the system should:
• Auto-validate obvious errors (e.g., missing ABN or mismatched totals)
• Rename the original file using a consistent pattern like “Supplier_Invoice#_Date.pdf”
• File it into a logical directory or feed an API/DB for downstream accounting
I’ll supply a representative invoice set for training and final acceptance testing. The finished deliverable should include source code, model weights (if applicable), a brief setup guide, and a simple way for me to add new supplier layouts without heavy re-coding.
Input formats are varied—PDFs, JPEG/PNG photos taken on phones, and scanned documents—so the pipeline has to be format-agnostic. My initial thought is a machine-learning-driven extractor that combines OCR with layout analysis, but I’m happy to hear alternative strategies if they boost reliability or speed.
Here is the data I must capture on every invoice:
• Supplier name
• ABN
• Invoice number
• PO number
• Invoice date
• Total amount and GST
• Line-item stock details (quantity, description, unit price or total value)
After extraction the system should:
• Auto-validate obvious errors (e.g., missing ABN or mismatched totals)
• Rename the original file using a consistent pattern like “Supplier_Invoice#_Date.pdf”
• File it into a logical directory or feed an API/DB for downstream accounting
I’ll supply a representative invoice set for training and final acceptance testing. The finished deliverable should include source code, model weights (if applicable), a brief setup guide, and a simple way for me to add new supplier layouts without heavy re-coding.
Apply on Freelancer →
Project sourced from Freelancer.com. Applications happen directly on the original platform — we never collect your data.