“I was sitting in the intro Computer Science course taught in Scheme when Lydia Kavraki came in and said, ‘I’m looking for some great students.’ She got our attention in our first semester, and I had a wonderful experience working in her lab for three and a half years,” said Rice University alumnus Jeff Phillips (B.S. in CS, B.A. in Mathematics ‘03)
Phillips is now an associate professor in the School of Computing at the University of Utah, and he credits his research experience in the Kavraki Lab for sparking his interest in academia.
He said, “Lydia took three of us, all undergraduate freshmen. Then she hired me as a summer research student. It was an unforgettable experience, that chance to start working in a research lab, to sit in on her group meetings. And she saw enough promise in me to bring me back.
“The next summer, we wrote a paper. Halfway through my undergraduate degree at Rice, I’d already taken the lead on a paper and become part of a respected research team. Through Lydia’s weekly group meetings, I began to see how all these unknown aspects were connected, and I learned how to approach interesting challenges.”
Phillips spent his third summer in an internship with a local company –a consultant with NASA– and he turned that experience turned into a paper as well. He returned to the Kavraki Lab for a final summer between his Rice graduation and the beginning of his PhD program at Duke University.
“It sounds like quite a trip, getting pulled into a research group and working on cool and interesting questions almost as soon as I arrived at Rice,” said Phillips. “But that experience is what led me into academia. I looked for internships as a freshman and, luckily, didn’t find any because I had this opportunity with Lydia that was much more interesting. Although, I might have made more money in industry.”
Phillips may joke about earning less as a professor, but his passion for research and teaching is a better fit with academia. He has been a key contributor to Utah’s emerging data science (DS) initiatives and helped create a graduate certificate program with an emphasis on DS for non-traditional students. His most recent contribution is the creation of a new bachelor of science degree in DS, launching this fall.
“About eight years ago, several faculty members teaching graduate classes in machine learning (ML), databases, data mining, and visualization all launched our courses about the same time. I was teaching data mining, and we began noticing a trend: an increasing number of undergraduates were trying to take the courses.
“So four years ago, we decided to do better job preparing the undergraduates for those courses. I introduced an undergraduate class that falls between statistics/linear algebra and the advanced courses in DS; it brings them up to speed mathematically and introduces ideas they will use in the advanced course. When they take the graduate level course, they will feel much more comfortable with those concepts. We realized we had the makings of a new degree – neither CS nor STAT – and started exploring it with a focus group of local data scientists. They said they wished the major had been around when they were still going to school. Two years later, it is being launched.”
His own evolving interest in DS is one of the reasons Phillips was so determined to create new DS opportunities for students at Utah. He said his tenure track role includes responsibility for teaching, research, and service. When his teaching role revealed a dearth of classes for the growing interest in DS, creating a major to match the demand seemed the best way to serve his students.
Phillips’ career path through academia has also taken some unexpected turns. He said, “I started out in robotics with Lydia, but wanted to attend grad school to better understand the more mathematical side of robotics. Instead, I ended up in a lab doing computational geometry. That drew on the broader aspect of the Kavraki group’s study of the geometry of proteins.
“After a few years, the specific problems I was working on didn’t seem they would make a significant impact on the field, based on feedback from biologists. As I considered other areas to explore with spatial data analysis, I began looking at data analysis from a geometric point of view and that has defined most of my work to date.”
Like many faculty members in CS and DS programs, Phillips is aware that most of his students who graduate will earn more than his professor’s salary. Some professors have already begun working in industry on research problems similar to Phillips’ with greater financial rewards. But Phillips recognizes his best work results from following his interests rather than income potential.
“I’ve realized the discrepancy between academic and industry salaries and decided not to pursue the money. There is something else that drives me,” he said. “As a faculty member, I am always busy with things I choose: interesting problems to work on, people to meet with, and other items I put on my own schedule. No one else determines my focus, there is no business concern to satisfy. I get to turn down or choose opportunities based on what I believe will have the most positive impact, and I choose the education of other people.”
He also values his freedom to determine how much time he wants to spend solving a problem. Phillips said his research problems are very interesting and he chooses them not because he has been instructed to find an answer for a customer, but because he’s looking or the simplest or best way to solve the problem.
“If I were doing this kind of research in industry, there would be a tradeoff to produce results. My perception is there is a kind of force, calculating effort against improvement. ‘If it takes more than a few days to figure out the right solution and that yields less than a 10% improvement, it may not be worth the time we’d spend on it.’ But in my world, I think it IS worth my time to simplify lots of science…my choice, my time.”
He said spending a lot of time to build up a new DS program was definitely not in his job description, but he envisioned the impact it could make for many people – not only the students but also the local business leaders and entrepreneurs who would go on to create the DS industry for the state.
“The goal wasn’t just creating something new, but also developing a clear and coherent explanation of the something new. I am working on a textbook for undergrads, explaining the mathematics of data science. I’m revisiting and rewriting algorithms and techniques, not to improve runtime but to achieve the simplest and cleanest explanation of it.”
Fortunately, Phillips has an abundance of energy to match his many interests. He said his best days are his busiest days.
“When I come in, and have all sorts of different things lined up, it’s probably going to be a good day. Like when I teach and nail the lecture, conveying a lot of useful information to students, then meet with my PhD students and talk through their research problems, maybe reach a break through. And that same day, I’ll have a meeting towards setting up the new program and we’re all in agreement on the path forward – it’s nice and clean and hasn’t taken up much time. Days when I have back-to-back challenges to meet and do so, that’s when I feel I’ve accomplished a lot.”