Portfolio
Description
Blinko is an iPadOS app teaching children vocabulary in a new language by identifying real-world objects via on-device machine learning.
It started from a simple question: what if kids could start picking up a new language as soon as possible by exploring the world around them? We looked into how young children learn vocabulary and found a study conducted with Rwandan school children learning French: those who learned by physically interacting with real objects performed 6 times better in vocabulary tests than those who just used pictures. That became the foundation of the whole app.
The app gives children missions to find everyday objects around them. They point the iPad camera at an object, CoreML recognises it on device, and the app shows an illustrated card with the word in the target language along with its pronunciation. Everything runs locally: no images are ever saved, stored, or sent anywhere.
After finding all the objects, two minigames reinforce what they learned. In the first one, children match illustrated cards to the corresponding words. In the second, they go searching again, but this time guided only by the sound of the word and a silhouette of the object.
The app supports five languages: English, Italian, Spanish, Portuguese, and Japanese. To validate the idea, we brought Blinko into classrooms and tested it with over 40 children in both public and private schools. The kids had fun looking for objects, repeating words out loud, and competing with each otherr. When they asked us to play again and again, we knew the concept worked.
Blinko was selected among the top 4 projects of the Apple Developer Academy (class of 2025) and presented to 500+ attendees including Apple engineers.