In Java, using Multi-Valued Decision Diagrams, or "MDD's"; Sequential Quadratic Programming, via IBM/Ilog CPlex; and, an implementation of convex hull theory, wrote a system to estimate anticipated parts demand to build vehicles still under development.

Estimates generated are used to anticipate required future plant capacity, and to negotiate with third party parts suppliers to identify which have the necessary capacity at the most competitive price. The majority of the heavy processing was run on Ford's High-Performance Computing environment.

The MDD's were used to represent the Families and Features available on a Vehicle Line and the Buildability of those Features with one another. A list of Features could be fed into the MDD and it would analyze them based upon the Rules from which the MDD was built to reflect whether the Features could be built together at the factory.

Later, we used the same MDD's to determine Features still available to a customer creating an online vehicle order, based upon the Features already selected by the customer.

At Ford, we followed the Agile development methodology.

We used Git for source control and JetBrains IntelliJ for software development; our applications initially ran on Pivotal Cloud Foundry ("PCF") virtual servers, we deployed using Jenkins, tested using JUnit 4, Mockito and other tools. The data was stored in a Microsoft SQL Server database. Later, we migrated from PCF to Google Cloud Platform ("GCP"), Tekton and PostgreSQL.