Project

We created two models to predict movies a user would like based on data from their Letterboxd account. The content-based model recommends similar movies to the currently displayed movie and the other model sorts a list of similar movies to a Users’ watched and rated Letterboxd reviews with a dataset. You can view a live website where the models are hosted here: MovieRecF24 A Letterboxd account is necessary to login, but no authentication is required to confirm you own the Letterboxd data. ...

We implemented a pretrained YOLO model from the paper You Only Look Once: Unified, Real-Time Object Detection which detects objects on a live camera. You can view the source code for our solution on GitHub.

We participated in the H&M Group’s Personalized Fashion Recommendations competition in April of 2022. We built our solution in PyTorch using an LSTM. You can view the source code for our solution on GitHub.