Data Science & Programming

Interactive tools, deep learning models, natural language processing pipelines, and data systems.

Interactive Tools & Software Prototypes

A completely client-side, browser-based environment for experimenting with, prompting, and observing conversations between multiple AI agents.

Interactive web tool helping clinicians and researchers compute Advanced Glaucoma Intervention Study (AGIS) visual field defect scores.

Interactive figures of merit calculators for clinical PPV, NPV, and screening cost-effectiveness trade-offs with downloadable models.

Interactive contour maps and geometric mean simulations demonstrating the hidden drag of volatility on long-term capital compounding.

End-to-end full-stack application featuring automated web scraping, neural article embeddings, personalized scoring, and web UI.

Multi-source search architecture orchestrating parallel retrieval across Reddit, Wikipedia, podcast transcript indices, and vector embeddings.

Applying deep learning and NLP to electronic health record clinical notes to predict rapid glaucoma progression and surgical intervention.

Published research in Journal of Biomedical Informatics evaluating clinical embeddings against domain analogies and prognosis prediction.

NLP classification models trained on PubMed abstracts predicting publication tier with 87.2% accuracy and 0.91 AUROC.

Fusing structured clinical tables with unstructured physician notes in deep neural networks to forecast low-vision progression.