After two years as a research scientist at Facebook, Rice University computer science Ph.D. alumna Yiting Xia ’18 began leading the network and cloud systems research group at Max Planck Institute for Informatics in Saarbrücken, Germany. “My vision for our future requires finding common consensus in the network community,” she said.
“I was inspired by Henning Schulzrinne’s recent keynote speech at SIGCOMM 2022 that examined network engineering from a new perspective —that of civil infrastructure engineering. Many people just see ‘the network’ only as an integral part of computing. But really, networking is the backbone of the communications infrastructure, and it is emerging like other civil infrastructures – water, energy, transportation – in the way it evolves rather than being replaced, because of how it is built over generations, and how its high initial installation cost is followed by even higher ongoing maintenance expenses.
“And now, the other civil infrastructures and the tools we depend on for research, medicine, transportation, financial trading, and activities of daily living are being built with machine learning, data parallel computing, and the Internet of Things (IoT) —all of which depend on networks. These tools are changing people’s lives and the networks that support all of them are growing.”
Xia said the building blocks of network architecture are already laid, but new areas of research are exploring the principles of how to make things work better. Issues like reliability and accessibility – similar to the importance of the electric grid’s reliability and accessibility – have become critical for networks.
“Demand for highly available, highly performing networks grows daily, which in turn brings cost into play. And increasing costs often mean decreased accessibility. That is a problem. We are also looking at the carbon footprint of our networks. Is the network footprint harmful to the environment and is it sustainable?
“My research revolves around those issues. Optical data centers can save costs and energy while providing high performance. It fits the future model very well,” she said.
Her work at Facebook focused on predicting and solving operational network problems, and she has continued collaborating with her former industry colleagues even as her career led her back to academia.
“We have been tackling real world challenges to network accessibility and reliability. We use metrics to measure reliability and resilience and metrics to quantify performance. We designed a system to mitigate these risks and when it is used, those key performance metrics improve,” Xia said.
She credits her ability to easily work in both industry and academia to her Rice advisor, Eugene Ng, and his emphasis on both quality research and the importance of communicating about that research clearly.
“Rice’s CS program topped the list of my Ph.D. admission offers because of its openness to a variety of backgrounds,” said Xia. “My undergraduate work was in electrical engineering and I was drawn to the intersection of computer and telecommunication networks. I wanted to pursue the computer science aspect of networks with my Ph.D. research, but most top tier research university CS programs require an undergraduate degree in that field.
“Even before I applied to Rice, I wrote to Professor Ng. He replied that he was recruiting graduate students for the next year and he was very encouraging —my lack of a CS degree would not be an issue. Then I arrived and experienced just how accepting Rice is of every student. It is an inclusive community no matter what your background is, and I think the rich diversity of experiences benefits the school in the long run.”
True to his word, Ng gave Xia space and time to learn and grow into her mastery of computer science. Xia said he never pushed her, never micromanaged her direction or her projects. “Eugene was a great advisor for my personality and my unique path, and that style of mentoring served me well,” she said.
But Ng did push his students to write and present at the very top of their game. He emphasized presentation skills to his graduate students, and had them give practice talks within their research group before every conference talk.
“Eugene gave very good feedback on how to improve our slides, how to actually give the talk, and how to handle questions at the end of the presentation,” said Xia. “I also took a research writing course in my early years at Rice that was very helpful in strengthening my skills in that area.
“Tracy Volz, the director of the engineering communication program, was instrumental in turning around a required seminar course, COMP 600, where CS grad students gave research talks every week. It was pretty bad in my first years at Rice. Then Tracy came in and gave a lecture about the principles for giving a technical presentation. She began working with each speaker one-on-one. They had to give a practice talk with Tracy and implement her feedback before they could present in the class.”
She met up with Volz again in Dan Wallach’s network security course. Wallach informed his students they would be required to give three presentations over the semester and he encouraged them to take advantage of the free communications improvement resources provided to Rice students by Volz. Xia practiced with Volz for all three of her presentations and said she was given very good feedback.
“When you go to work in industry, you can clearly identify the Ph.D. graduates in your area because they usually have much better communication skills,” said Xia. “In the industry, not only do you have to write and give technical status updates, you also have to give talks and write technical posts.”
“You have to share new and changing information with your team members, and convince your manager to go with your plan. That will require you to deliver your ideas clearly and succinctly and win your stakeholders’ trust. Rice’s Ph.D. graduates have no problem communicating in industry and in technical discussions because we are super used to it from our research group conversations.”
She has a different view of a typical Ph.D. student’s preparedness for communicating in academia and said proposal writing is not like writing research papers for journals.
“In academia, your research proposals will require you to collaborate with people from even more diverse backgrounds than you experienced in your university. You will probably be working with professors who got their Ph.D.s in different countries and have very different working and communication styles.
“The way you present also shifts. Even if you mastered the art of giving conference talks, your audience in academia is constantly changing. You’ll give talks to faculty; some of the professors in the audience will have deep experience in your area and others will not. You’ll give tutorials to students, and you’ll give very entry-level talks about your topic to the general public,” she said.
In addition to organizing the content of a talk differently for each audience, Xia also recommends looking for a mentor in each new environment. She said younger faculty members are often matched with senior professors but they can also seek out additional mentors.
“No one flies on their own,” said Xia. “Talk to your colleagues as well as other professors. Ask their advice, but also figure things out as you learn from your mistakes. Maybe an audience doesn’t respond well to part of your talk. Change it, continue to iterate, and learn.”