Document AI

Testing Financial Table Extraction on Exchange Daily Records

Large-sample validation and defect-driven hardening of Daily Record extraction.

PythonDoclingjq
Project name
Exchange Daily Record – Large-Sample QA & 99.99% Accuracy Hardening
Company
Strategic Systems International
Role
Principal Software Engineer – Lead Developer
Illustration of Testing Financial Table Extraction on Exchange Daily Records

Project overview

I built a validation testbed for a stock exchange’s Daily Record publication. Its jq checks look for specific failures in extracted JSON, including merged rows, missing first rows, tickers carried into the wrong record and dividend tables split in two. I ran the checks on a large delivered sample to guide work toward the project’s 99.99% accuracy target.

Tools & Technologies

Python · Docling · jq · SQL · MySQL · automated validation testbed

Project illustrations use sample data.

Need a similar system?

Let’s discuss how I can help build a solution around your documents, data, and workflows.

Discuss Your Project