I will extract data from pdfs and invoices into excel or json with ai
Backend and AI fixes for Supabase, Stripe and OpenAI apps
About this Gig
Your team copies numbers out of PDFs by hand: invoices, bank statements, purchase orders, forms. It is slow, and one typo costs money.
I build AI extraction that does the copying. At my day job I am the main backend developer on a platform where AI agents pull 300+ fields per company out of annual reports, and analysts check every value against its source page. The same approach works on your documents.
What you get:
- The fields you need, pulled from PDFs, scans or images
- Output in Excel, CSV, JSON or straight into your database
- Each value linked to its page, so checking is fast
- Flags on values the AI is unsure about
- A script or small API you can run again next month
Not included: manual data entry, handwritten notes, or ESG and sustainability reports.
Send me 2 or 3 sample files and the list of fields you need, and I will tell you what accuracy to expect.
Technology:
Excel
•
Google Sheets
•
Python
My Portfolio
FAQ
Can it read scanned PDFs and photos?
Yes. Scans and photos go through OCR first. Very poor scans lower accuracy, and the Sample run shows you the real accuracy on your own files before you order a bigger batch.
Is my data safe?
Your files are used only for your order and deleted after delivery. For sensitive documents, I can run the extraction with the AI provider you choose or on your own server.
How accurate is it?
It depends on your documents, so I do not promise a number before seeing them. Every value comes with its page so you can spot-check fast, and unsure values are flagged for review.
When do you work on my files?
Evenings Pakistan time, which is the US morning and early afternoon. Messages get a reply the same day, and the delivery time on each package already allows for this.

