Nuno Veiga

Our customer has recently acquired a new billing system to replace their legacy one and wanted to validate if the new system would act like the old one when generating invoices.

Project Steps:
01. Loading the as-is data from their old system into our application.
02. Loading the as-is data from their new system into our application.
03. Performing ETL processes on the data from step 01 and checking for any data discrepancies.

We would load three months worth of data at one time and once the customer validated that the applied transformation was correct, the data was wiped out, and another set was loaded.

Our customer had recently acquired a new billing system to replace their legacy one. In order to validate that the customer invoices were being properly made, we created a system that retrieved the data directly from the customers' Exadata Oracle databases.

The process consisted of reading the data from the Exadata into CSV files and transferring them to our data center via a secure copy protocol (SCP). Once the files reached our server, we would perform an ETL (extract, transform, load) over that data to feed the data validation processes and reports. All of these processes ran in near real time.

I worked on several projects that consisted of loading call-detail records (CDRs) into our system in order to detect fraud patterns and/or revenue leakage. Most of the activities consisted of ETL flows to load and transform data from several sources and use them for reports.