Entry
A lot of students encounter machine learning as a toolbox of recipes, while analysis and probability feel like separate classes they must survive before getting to the practical material. That split is a teaching problem, not a student problem.
In workshops, notebooks help collapse the distance between theorem and application. You can show the idea, run the example, change a parameter, and watch the behavior move in real time. That makes the motivation operational instead of abstract.
The curriculum work I care about is not just about slides or demos. It is about sequencing ideas so that linear algebra, probability, optimization, and signal processing reinforce each other. When that happens, students stop seeing theory as decoration and start using it to reason about failure modes and model behavior.
At UF, that teaching work also feeds back into my research. Explaining concepts clearly forces a higher standard on the pipeline design, the examples I choose, and the way I think about reproducibility.