Jason Senthil

Bachelor's

Computer Science

Software Engineering

Seagate

Machine Learning Intern

- Developed a solution for time-series unsupervised anomaly detection on sensor data to predict tool failures in the hard drive factory, utilizing gradient boosting trees and robust statistics - Trained and deployed Convolutional Autoencoders conditioned on a Feature Pyramid Network's (FCNs) output for - Researched deep learning approaches to unsupervised anomaly detection in time-series, used Fully Convolutional - Networks for varying sequence lengths, and Soft-Dynamic Time Warping as a differentiable loss function