Created PyTorch Geometric Siamese-GNN to predict financial crises contained within the International Trade Network.
Created latent space model to predict co-evolution of trade, migration, and terrorism networks, achieved AUC of 0.91. Accepted to Networks and Time II Conference hosted by Northeastern University in London.
Implemented traditional and deep learning ML models for classifying forms of advocacy contained in 21 million BLM tweets. Analyzed model classifications, discovering shift away from within-the-system and towards disruptive forms of activism.