SVM Eye-Color Classifier — Curoverse / Veritas Genetics

Built and published a support-vector-machine classifier predicting eye color from genomic data with 95% accuracy. Presented at the Harvard I2B2 TranSMART Symposium.

Publication · Saldana, J. — Eye-Color Prediction via SVM on Genomic Markers

Bioinformatics data-science internship at Curoverse (later acquired by Veritas Genetics). Built and published a support-vector-machine classifier predicting eye color from genomic markers with 95% accuracy, presented at the Harvard I2B2 TranSMART Symposium.

The accuracy figure isn’t novel research — eye color is among the better-understood phenotypes — but the experience of going from raw genomic data through model selection, validation, and a published technical artifact was foundational. It was the first time I shipped something that lived outside the codebase that produced it: a model, a paper, and a presentation, each of which had to stand on its own.

The publication has a permanent DOI: 10.5281/zenodo.1045265.