Even for an "engineering-heavy" project where the main deliverable is a set of tools or software package, there should be an evaluation component.
For example, below is some phrasing from an MQP that built an AB testing framework:
Each portion of the framework was tested individually. The Rust Crate and REST API were both unit tested, while the UI was tested manually. The platform was system tested with a demonstration. The demonstration objectively shows that when combined with the sponsor's software, the framework provides working functionality for A/B testing, and conversion events can be tracked. Based on our tests, demonstration, and the expectations of our sponsors, the framework is a successful minimum viable product for an A/B testing framework. Overall, our system provides the foundation for future experimentation in a serverless space and shows a future for products.