AI Builders: High School Students Train Their Own Image Classifier at Columbia

This summer, Aiphabet partnered with SHAPE, Columbia Engineering’s high school outreach program, to teach “AI Builders: From Algorithms to Applications.” Over three weeks, 27 high school students went from AI users to AI creators, working through Aiphabet’s platform lessons on machine learning, decision trees, k-nearest neighbors, perceptrons and neural networks, alongside Python notebooks.

The course ended with a hands-on project: students photographed landmarks around Columbia’s campus and used transfer learning with a pretrained ResNet to train their own landmark classifier. Along the way they tested data augmentation, read confusion matrices, and learned why models overfit. Teams presented their work at a final Demo Day.