All that Amarda Shehu (Ph.D. ’09) knew when she chose Rice University over other graduate schools was that Houston was warm and that Lydia Kavraki’s research sounded remarkably interesting.
“I came to Rice in 2002,” said the associate professor of computer science at George Mason University. “I had spent my undergraduate years in a lovely engineering university in upstate New York, but I was fed up with the snow and cold. I drew a line at the mid-Atlantic U.S. and only applied to grad schools below that line.
“I added an application to Rice almost at the last minute, because I noticed it had one of the top ten computer science programs, and, of course, it met my climate requirements,” she said.
After visiting several campuses, she said, “I arrived at Rice in what must have been late February or early March. The weather was wonderful!” Shehu said. “Rice felt like Stanford, only smaller and more intimate. I talked with several professors and was impressed. I also got to talk with current graduate students and even the administrative assistants in the department.”
During her visit, Shehu assumed she would join a theoretical computer science group. Her plans changed: “I recall sitting there and listening to short research highlights from faculty. One of them was Lydia. The moment Lydia started showing molecular simulations, everything changed for me. As soon as I started seeing those atoms move around, I turned to my boyfriend at the time, who had also been accepted at Rice, and said, ‘I changed my mind; that’s what I want to do.’ I had no clear idea of what I was looking at, but that intersection of biology and computer science in Lydia’s research intrigued me.”
Shehu joined Kavraki’s research group and was given her first project.
“Lydia assigned me a small task on molecular structure visualization,” she said. “It was an opportunity to learn the terminology of the domain and pursue something manageable of my own in the context of a larger project. Kavraki sets her new graduate students up for success by giving them small parts of a larger group project, and the first-year students fall under the mentorship of more advanced students.
“As soon as I got comfortable, Lydia assigned me my own project on modeling molecular structural flexibility. It was great but challenging. It required me to learn about inverse kinematics, molecular geometry, and thermodynamics. Progress was slow at the beginning, but Lydia believed in me and the project, and encouraged me to apply for an NIH training fellowship, the Keck fellowship, from the Gulf Coast Consortia.”
The application involved pitching her research idea in a short presentation to a three-member committee, and Shehu was intimidated. Beside her were Kavraki and her collaborator, Cecilia Clementi from the chemistry department, who became Shehu’s co-adviser on the project.
The fellowship encouraged training by two advisers in different fields. “Looking back at it, I’m sure Lydia and Cecilia were nervous during my presentation, but they were ecstatic afterwards.”
Shehu’s project was modeling the equilibrium dynamics of proteins. “We started simple, focusing on modeling first loop motions in proteins. As our capabilities grew, our projects became more ambitious,” she said. It became her M.S. thesis and resulted in two high-impact journal publications.
Shehu says, “It was difficult to please two masters. I had so much catching up to do in two different fields with their own languages and own standards. I took lots of courses outside computer science.”
Kavraki recognized her struggle. “Lydia got me out of the office one day and drove us to a nice little sandwich shop. I was nervous hanging out with my adviser in a non-work setting. I don’t recall all our conversation. One thing I do remember is Lydia telling me I was a computer scientist, and my priority had to be novel algorithmics.”
Shehu remembers similar mentoring occasions when returning with Kavraki from off-campus events such as seminars and symposia. “I realized Lydia was investing in me, and at the same time pushing me to take control of my own work.”
Once Shehu realized she had autonomy over the direction of her doctoral dissertation, she began to enjoy herself. “It was completely my baby. I would prepare these detailed, paper-style progress reports every week, and present them to Lydia and Cecilia. They would give me feedback, redirection at times, and words of encouragement. Whether it failed or succeeded, it was mine.”
In preparation for a conference where Shehu would present to an audience of several hundred scientists, Kavraki and Clementi made her go through rehearsals in front of members of their groups. Kavraki instructed them to play the role of a hostile audience and ask questions that might expose shortcomings in the work. “She wanted to see how I would react, and how I would answer in an environment that scared me.”
Shehu has drawn from such experiences in training her own graduate students: “I need them to understand that I was once a student, too, vulnerable at times and terrified. But I also need them to understand that this is part of growing up, and I credit Lydia with sharing this with me.
“The adviser-student relationship is a deep one,” Shehu said. “It never goes away, and I consider myself lucky that I found a great adviser who not only welcomed me in her group but invested in me and stuck with me.”
Shehu received tenure at GMU in 2014. “There is before-tenure and after-tenure success,” she said when asked about her high points. Before tenure, Shehu considered it a good day when she received news of a paper or grant proposal being accepted. Now she realizes that these are instant gratifications.
“I try not to make too much of such days,” she said, “because research has its own ups and downs, sometimes many downs.”
Shehu tries to emulate her adviser and maintain a positive attitude: “I have learned that even bad days can lead to good ones. A really good day is when I learn something from my students. When they tell me something I did not know, when they are actually feeding me, and I realize they are starting to fly on their own. That is a really good day.”
Amarda Shehu completed her Ph.D. in CS in 2009. Her adviser was Lydia Kavraki.