A Convolutional Recurrent Neural Network (CRNN) and audio processing pipeline for detecting choruses in music. Trained on a custom dataset of 332 annoted songs (manually labelled by yours truly) and achieved 89% accuracy on a holdout test set.

A question-answering system built using the multi-qa-distilbert-cos-v1 sentence-transformer model, fine-tuned on a custom dataset to help customers find information about various medical equipment.

This project explores the application of Chromagram Self-Similarity Matrices (SSMs) for segmenting music into its structural components such as intro, verse, chorus, and outro.

Unsupervised method of segmenting music using various audio features and novelty/peak detection.

Developed a decision tree model to predict customer purchases for a specialty coffee company, analyzing factors influencing buying behavior and providing actionable insights for marketing strategies

Collection of case studies I completed that cover a wide range of industry-relevant problems, such as customer segmentation and purchasing predictions, using techniques like clustering, decision trees, and regression to extract business-critical insights.

A DJ and mixtape generator that uses deep learning, Python, and a sprinkle of music theory. (This is an ongoing project)