NCR: Amtrak Point-of-Sale Implementation

Providing Enterprise Architecture consulting in a very large scale point-of-sale implementation

Client: NCR and Amtrak


For NCR’s point-of-sale implementation for Amtrak, the project was a multi-year effort spanning many vendors and technology resources across the globe and had a budget in excess of $13M. We provided consulting and development support to help executives get an understanding of the technology landscape. We recommended and implemented streamlined cross-vendor development processes, including shared source control and various release management/code review strategies. We also provided development support for the core point-of-sale platform and various integration components that support inventory reporting.

Technologies and Tools

Enterprise Architecture, Business Analysis, C++, C, Visual Basic .NET, SQL Server, XML, Subversion


Everything about this project was astounding, in every sense of the word. First, NCR is a corporation (their market cap was over $5B at the time). Second, their client, Amtrak has very publicly been fraught with all sorts of challenges. And this was just the beginning. We joined the project to help provide some “sense” to what was going on from a technical perspective, but later discovered a wide range of severe technical and political issues. This project became the ultimate test of our mental fortitude as well as our technical aptitude.

So we had to quickly get a handle on the landscape politically and technically, which was an exciting challenge. We soon had a slew of point-of-sale tablets in the IntegrityTech HQ and were back-and-forth between Georgia Southern University in Statesboro, GA (where one of the prime vendors was located), NCR’s office in Duluth, GA and Amtrak’s HQ at Union Station in Washington, DC.

On top of that, the primary core architectural issue was that the system on the trains had to operate disconnected from the internet, yet be able to process credit card transactions and interact with a remote inventory system. The secondary core architectural issue was that the main point-of-sale terminal software was a port of the original terminal (green screen) to Windows and used a graphical emulator for placing items on the screen (no WYSIWYG here).

The cool part? Well, we’d like to say that we somehow managed to “save the day”, but that just isn’t the case. We did manage to bring a lot of people together who had previously refused to work together and communicated the architecture in an understandable way to the executives at NCR and Amtrak. We’d definitely consider that a win.