Jose Javier Gonzalez Ortiz

  • PhD Student, MIT
I am a PhD student at MIT CSAIL in the CAML group, advised by John Guttag. I’m interested in efficient machine learning methods, investigating how to make current machine learning more computationally efficient and how to reduce training costs from an algorithmic and model design point of view. Recently I’ve been exploring HyperNetworks as a way to perform amortized model learning, substantially reducing the computational cost of hyperparameter tuning.
Previously, I’ve worked on pruning methods and distributed training for deep neural networks. While interning at Facebook AI Research Montreal I studied how parallel SGD methods scaled to large image classification tasks and improved current accuracy-communication trade-offs. I developed ShrinkBench, a library to facilitate development and standardized evaluation of neural network pruning methods. In the past, I have worked in machine learning and computer vision for healthcare applications.
At MIT, I have also had the opportunity to teach several classes. I co-organized Missing Semester, a lecture series on programmer tools often left untaught in modern CS curriculums. I was also a TA for MIT’s graduate distributed systems class 6.824 as well as Intro to Deep Learning 6.S191. Before joining MIT, I received a BSc in Telematics Engineering from ICAI in Spain. During my undergrad I interned at CERN Openlab and studied for one year at the University of Michigan, Ann Arbor under an exchange program.


Awards & Honors