Built prediction engine for text classification. Created several thousand data elements to train the model. Improved prediction accuracy from 65% to 94%. Analyzed options for text classification such as a RandomForestClassifier, SDGClassifier, and Linear SVC. Majority of project was done through python.

We created a timer app directed towards children on the spectrum who have trouble staying on task. Using React Native, Firebase, JavaScript, and Expo-Speech the application is designed for a supervisor to create a set of tasks for the child to help them complete regular tasks such as brushing their teeth. The child can be given tasks that show images while also playing audio of the task at the time. This helps them retain attention to a specific task using by aiding their visual and auditory senses.

The school uses visual and sensory aids to make learning substantially easier and intuitive for the students. One of the methods the teachers use is word association with pictures. The students move images around to create word associations between subjects, objects, and verbs. The teachers use this method as kids on the autism spectrum sometimes struggle with paying attention. The teachers wanted an application that would be a digital version of the monolithic booklets. The students would then use desktop computers and tablet devices to do the exercises. The app made it easy for teachers to create and customize their lessons.