Teaching and mentoring

Students struggle when the mental model they have built differs from the one the problem requires, and the code is only where the difference becomes visible. When a student brings me a program that does not work, I ask them to explain what should happen at each step, and the gap usually appears in that explanation.

I build courses around a cycle of building, measuring and explaining. Students construct something that works, measure how it behaves, and account for the choices they made. Because generative AI makes a submitted file weaker evidence of learning, I assess reasoning alongside output, through short code reviews, prediction tasks, and exercises in which AI-generated code becomes the object of study.