In this discussion, I am looking for ideas on how we can improve project course teaching to produce more industry-ready graduates. The University of Canterbury offers a third-year, two-semester, group project course that has been available since 2012. This course provides an opportunity to experience firsthand the joys and sorrows of working together as a team to develop a large, complex software application. Project development is based on the SCRUM framework. The course focuses heavily on "learning-by-doing," collocation, and knowledge sharing. Every year, we have 80-90 students, which forms 10-12 teams. Some of the challenges we are facing relate to student adoption of quality assurance practices. Initially, they have a basic understanding of unit testing and find it relatively easy to incorporate it into their development practice. More advanced development practices, like BDD and ATDD, remain a challenge for most. They are seen as overhead and effort-prone, so teams often revert to manual acceptance testing practices. There is an increase in AI-produced code in the project codebase, and we are trying to understand how this tendency is affecting learning outcomes. Also, it would be interesting to know what other skills graduates need to have to adjust quicker to industry demands.