Build and train a small model, then find out where it fails and why that matters.
Not a tour of chatbots. Students train a classifier on data they collect themselves, watch it get things confidently wrong, and work out why. Ends with each student writing an honest assessment of a tool they already use daily.
Dina Farouk
ML engineer, 9 years in production machine learning
Builds recommendation systems for a living. Teaches this course because most teenagers are already heavy users of systems nobody has explained to them.
“I thought I understood this from videos. The first time my model was 94% accurate and still useless, I actually got it.”
Hana M.
Student, age 16
No. Students train models and look at the results. Prompting is covered in one session as a tool, not as the subject.
That is fine. We provide working notebooks and teach the Python needed along the way.