Over the past decade, I have implemented flipped classroom formats in first-year mathematics courses for engineering students. This talk presents a retrospective view on this experience, combining active learning practices with reflection on teaching during the critical transition to university-level STEM education. The presentation reports results collected over multiple course offerings indicating that flipped classroom formats are associated with more inclusive outcomes in first-year mathematics. In particular, performance gaps related to students’ prior educational background and gender are reduced in flipped classes compared to more traditional instructional formats. These findings suggest that active learning environments can better support a diverse student population in large first-year cohorts. Beyond these results, the talk focuses on lessons learned from ten years of teaching flipped classes. Attention is given to the practical challenges of sustaining active learning at scale, as well as to the limits of flipped instruction in mathematics. Rather than promoting a single instructional model, the presentation reflects on which elements proved robust over time, which required continuous adjustment, and which initial assumptions needed to be reconsidered. The talk concludes with a personal perspective on flipping first-year mathematics classes, highlighting insights that may be relevant for teaching computational science and engineering across undergraduate and graduate levels.