We describe a collaboration between Marelli and Università degli Studi di Milano that allowed the latter to add a course on «Architectures for Big Data» in its Master programme of Computer Science, with the aim of providing a teaching approach characterized by an intertwined exposition of discipline, methodology and practical tools. We were motivated by the need of filling, at least in part, the gap between the expectation of employers and the competences acquired by students. Indeed, several big-data-related tools and patterns of widespread use in working environments are seldom taught in the academic context. The course also allowed to expose students to company-related processes and topics. So far, the course has been taught for two editions, and a third one is currently ongoing. Using both a quantitative and a qualitative approach, we show that students appreciated this new form of learning activities, in terms of enrollments, exam marks, and activated external theses. We also exploited the received feedback in order to slightly modify the content and the structure of the course.