What is this page for?
โ† Selected work

Workflow automation ยท Perrigo

From days of reporting to minutes

Python, OCR and layout parsing across three ERP networks, replacing a manual reporting process with automated reconciliation.

The operational problem

At Perrigo, financial reporting and reconciliation crossed three ERP networks. The workflow required people to capture information, extract it from documents and reconcile discrepancies before producing reports.

My contribution

I built Python automation for SAP invoice auditing and reconciliation, using Tesseract OCR and layout parsing to detect unit-of-measure discrepancies. I also built reporting pipelines covering screenshot capture, PDF extraction and text cleaning across the three networks.

Outcome

My professional history records a reduction from a multi-day manual reporting process to minutes. This is a reported work outcome; the portfolio does not expose client records or a timing dataset.

Why it belongs in an FDE portfolio

This work starts with the existing workflow and connects systems that were not designed as one application. It demonstrates integration, handling messy inputs and delivering a useful operational result, alongside the model-based systems elsewhere in my portfolio.

Source: Perrigo in my professional history.

Discuss this work โ†’

๐Ÿค– What this page is for

Read a public case study with the problem, engineering contribution, evidence and limits, or try the protein structure visualization without signing in.