Dataset optimization and control

On day one of a project, a manager or customer will inevitably ask the AI expert what kind of data is being dealt with, and in which format. Providing an answer to that will take some hard work. The AI expert never really thought of that, having acquired expertise using the available, optimized, and downloadable training datasets to test the programs. The expert's name is Pert Ex (an anagram of expert) in this chapter. The expert likes to be called Pert.

The CEO of the corporation (or boss or customer) wants Pert to lead an Amazonization project. Pert doesn't quite understand what that means. The CEO has the patience to explain that wholesale e-commerce has overtaken traditional ways of doing business. Wholesale means increasing use of warehouses. How those warehouses are managed will make a difference to the cost of each project and the sales price.

Although the business is now huge and the warehouse is expanding, the costs of retrieving products from the warehouse locations have been increasing. The corporation, which we will name AGV-AI in this chapter, represents the sum of several real-life projects I have managed on various warehouse and aerospace sites. Bear in mind that moving large components in an aerospace project requires precise scheduling. Moving large volumes of small products in an Amazonization process or small volumes of oversized products (25 meters or 82 feet for example) such as aerospace components can be costly if badly planned. In AGV-AI, automatic guided vehicles (AGVs) are surprisingly taking longer routes from a warehouse location to a pier. AGV-AI manufactures, stores, and delivers large volumes of consumer goods.

The CEO wants Pert's team to find out what is going on and Pert must lead the technical aspects of the project.