From 430 MB to 13 MB: Building an Offline AI Model for Indian GST Invoices

A developer has released Tiny GST, an on-device AI model that extracts structured data from Indian GST invoices, shrinking the model from 430 MB to 13 MB. The project aims to cut cloud costs, keep sensitive invoice data local, and enable invoice processing in warehouses or rural areas with unreliable internet. The release is early-stage: its XML workflow has been tested only against a mock Tally server, with live TallyPrime import testing still pending.

We built Tiny GST, a compact AI model designed to extract structured data from Indian GST invoices directly on-device. The model: The goal is simple: reduce cloud costs, protect sensitive invoice data, and make invoice processing work even in warehouses or rural areas with unreliable internet. The project is currently an early developer release. The XML workflow has been tested against a mock Tally server; live TallyPrime import testing is still pending. 🔗 GitHub profile: https://github.com/sanprat https://github.com/sanprat 🔗 Tiny GST repository: https://github.com/sanprat/GstEngine https://github.com/sanprat/GstEngine 🔗 Full engineering article: https://pybankers.com/blog/from-430mb-to-13mb-offline-ai-gst-invoice https://pybankers.com/blog/from-430mb-to-13mb-offline-ai-gst-invoices 🔗 PyBankers: https://pybankers.com https://pybankers.com