UG's Common Decision Engine (CDE) provides a means of automatically making mortgage insurance decisions for the company without the intervention of a human underwriter. In the past UG had used expensive and complex systems to perform this task. These systems differed for each and every mortage origination system at United Guaranty: at least four separate decision engines in total. This caused repeated problems for the company when the origination systems were not in agreement (a result of the various systems becoming out of sync over time). Also, the legacy systems were hardcoded in legacy languages which neither fit into the new eBusiness architecture nor allowed the business units the direct control of business decisions. Because of these legacy systems and the need to code, test, and promote any business logic changes, UG was relatively slow to respond to changing business needs.
The CDE provides a data-driven service that accepts a loan application object, performs calculations and data lookups, and utilizes the Business Rules Engine to apply business logic. The CDE analyzes the failed rules and determines whether the loan should pass, fail, or be referred to a human underwriter.
My involvement with the CDE was from inception to implementation. I did the primary analysis for the project, conceived of and documented the business design, and did the Functional System Design. I took the lead in presenting this design to UG's Architectural Design Team and incorporating their suggested changes into the design. As the project passed into the construction phase I worked with the programmers to ensure that the implementation coincided not only with the planned design, but that it provided flexibility to interface smoothly with both our eBusiness architecture and legacy architectures. As the project began to be incorporated into existing systems, such as EDI, I again took the lead in working with these other teams to iron out the interfaces and establish responsibilities to enable a smooth transition to the new architecture.
The CDE incorporates and extends prior infrastructure projects such as the Business Rules Engine and the Field Validation Engine as well as other existing services such as a data access abstraction layer and existing legacy pricing and scoring components. In addition to making the most of these prior investments, this decision resulted in a design that is clean and very simple when seen at a high level. It is flexible enough to be easily extended, and modular enough that as services are improved (as with a new pricing component) the improvements can be implemented with minimal effort.