ERP Migration

As part of the Tuff Shed engagement, the team also initiated a large-scale data cleaning and migration effort to support an ERP implementation with Nextworld. The migration scripts — a collection of Python Jupyter notebooks running on Azure ML compute instances — handled the ETL pipeline for moving fiscal and operational data from JD Edwards into the new system. The work involved reading CSVs from Azure Blob Storage, handling legacy cp1252 character encoding, cleaning over a million GL transaction records across five partitioned files, and processing dimension tables for chart of accounts, suppliers, serial tags, and company structure.

The DataMigrationService in the core monorepo (28 controllers, 40 orchestration classes, and 18 embedded CSV seed files totaling over 1.1 million records) was built to ingest and transform that cleaned data. The JD Edwards-to-Nextworld migration was ultimately canceled when the business relationship with Nextworld soured — but the DataMigrationService remains in place today, serving as the broader data infrastructure for ongoing migration and integration needs across the platform.