Developed and incorporated an interpretability module for machine learning models processing over 800 million data points across more than 220 schools in over 10 countries.

Integrated the Python and R back-end for the analytics app, allowing seamless use of models developed in both languages

Trained SVMs, Random Forests, and Neural Networks to predict customer credit risk from a FICO home credit dataset. Generated instance explanations for the models based on feature importance.

Won Second Place in the FICO Explainable Machine Learning Challenge

Performed dimensionality reduction with deep autoencoders on the spectrogram of the audio and the extracted MFCCs. Developed an interactive webapp to explore musical similarity.

Led the computational analysis for an Arab/African music collection with more than 10,000 songs.