The mysteries and fun of chemistry lured Jose A. Garrido Torres to devote his academic career to the discipline. But as he pursued the subject, Garrido Torres still wanted to maintain his interest in computers, particularly the application of computing to data science. In his academic career, Garrido Torres combined these two disciplines and now specializes in an exciting subset in chemistry: applying advanced machine learning tools to optimize reactions and make new and useful compounds.
"It's an exciting time to be in chemistry because machine learning and artificial intelligence are opening new avenues of discovery and doing so at a quicker pace than what was possible before," said Garrido Torres. "This recent interest in using machine learning in chemistry has occurred in the last six or seven years, so this interdisciplinary topic has plenty of room to grow."
Garrido Torres, a data scientist specializing in chemical catalysis, joined Princeton University in November 2020 with support from the Schmidt DataX Fund. The fund aims to spread and deepen artificial intelligence and machine learning across campus to speed scientific discovery.
Princeton announced this new fund in February 2019, a gift of Schmidt Futures, with the Center for Statistics and Machine Learning (CSML) overseeing part of this initiative. This fund includes the support of six Schmidt Data Scientists - Garrido Torres being one of them.
The data scientists' mission is to help drive the development and improvement of data-analyses to enable more impactful research in three areas on campus: the Princeton Catalysis Initiative, the biomedical data science initiative, and the Center for Information Technology Policy (CITP).
Garrido Torres works for the catalysis group. His specialization is in computational chemistry with expertise in machine learning, artificial intelligence, Bayesian optimization, surrogate models, atomic-scale simulations, scripting and programming, and other topics.
"We are excited that Jose is part of the DataX team," said Peter Ramadge, CSML director. "His experience and expertise in chemistry and data science is an asset to the catalysis group who are doing great, exciting work in accelerating and pushing innovation in chemistry."
Abigail Doyle, the A. Barton Hepburn Professor of Chemistry, has been a significant driver within the catalysis group. Doyle had recently published a paper, "Bayesian Reaction Optimization as a Tool for Chemical Synthesis," in Nature in February 2021 in collaboration with Ryan Adams, professor of computer science and director of CSML's undergraduate certificate program. DataX supported this project with seed funding.
Before coming to Princeton, Garrido Torres worked on an open-source, machine learning-driven platform for bench chemists to optimize chemical reactions. This work has similarities with the research in the Doyle-Adams project. As a data scientist on campus, Garrido Torres will continue to expand this machine learning approach to optimize chemical reactions.
Garrido Torres is also working on a project that involves building datasets for certain classes of compounds and reactions.
"What I like to focus on is creating frameworks that allow us to explore the chemical space autonomously. I build models to help these explorations and to make predictions," said Garrido Torres. "And from these explorations, we hope to speed up the search for new chemical compounds for use in medicine and other fields."
Garrido Torres majored in chemistry at the University of Alicante in Spain, where he graduated in 2012 with a bachelor's degree. He stayed at the same school for a masters' degree in materials science, which he earned in 2013. His research focus then was modeling in materials science.
He then went to the University of St Andrews in the United Kingdom to study for a doctoral degree in computational chemistry and materials science, which he earned in 2017. His thesis' title was "Density Functional Theory investigations of molecules on surfaces: from nano-electronics to catalysis." During his doctoral study, he was also a visiting research scholar at the University of Vienna in Austria.
After getting his doctoral degree, Garrido Torres became a postdoctoral researcher at Stanford University. His focus area was artificial intelligence applied to accelerate atomic-scale simulations for catalysis. He developed CatLearn, a platform for building and testing atomic machine learning models.
Before coming to Princeton, he was a postdoctoral researcher at Columbia University from 2019 until his appointment to DataX. At Columbia, he developed machine learning methods for the modeling of high-temperature electrochemistry for metal recycling.
His academic contributions include numerous seminars and presentations, in addition to 14 papers published in journals.
Garrido Torres regards the camaraderie and the cooperation among experimentalists and theoreticians as very positive aspects of his position.
"I am very excited to be here," said Garrido Torres. "The caliber of the work at Princeton is fantastic. And I get to interact and work with great people who are passionate about their research. I feel lucky that I have a role to play here."