UC Berkeley AI Hackathon
Developed an AI-powered mobile app for clothing recommendations.
Portfolio
AI Personal Trainer Research Paper
Research abstract: Regular physical exercise is crucial for health and wellness, but performing exercise with incorrect form can cause injuries. This paper presents a machine learning-based method to classify the quality of squats. Using an iPhone to record exercise, the videos are processed using Google’s Mediapipe, which extracts x, y, z coordinates of body joints. These coordinates are then converted into joint angles, in particular, the knees and hips. A deep neural network analyzes these angles to classify and predict the quality of the squat form, offering a solution to improve exercise safety and effectiveness. 
Car Price Prediction
Given Facebook marketplace data, I used supervised machine learning to predict the prices of cars in the Bay Area, as a function of features such as: mileage, condition, and seller rating. 