For a better experience, click the Compatibility Mode icon above to turn off Compatibility Mode, which is only for viewing older websites.

CCI Faculty Spotlight: Vasilis Gkatzelis

April 17, 2017

Vasilis GkatzellisVasilis Gkatzelis, PhD is an assistant professor of computer science in the College of Computing & Informatics (CCI). Before joining Drexel's faculty in fall 2016, Gkatzelis served as a postdoctoral scholar at the University of California, Berkeley's Department of Electrical Engineering and Computer Sciences; at UC Berkeley's International Computer Science Institute and Simons Institute; and at Stanford University.
CCI: Tell us more about your research interests. What are some of the real-world applications of algorithmic mechanism design?
Vasilis Gkatzelis: My research lies at the intersection of computer science, economics and operations research. Most of my work focuses on the design of fair and efficient mechanisms that allocate valuable resources among strategic agents. For example, consider the ongoing Federal Communications Commission (FCC) spectrum auction. With a strong demand for wireless connectivity and the proliferation of data-hungry smartphones, the FCC spectrum auction aims to redistribute the licenses for transmitting over the electromagnetic spectrum. The agents competing in this auction are large telecommunications companies like AT&T, Verizon and T-Mobile who would, of course, be willing to misreport their value for these licenses, if this led to lower prices. The goal of algorithmic mechanism design in this setting is the design of computationally efficient auction mechanisms that incentivize the participating agents to behave in a way that leads to efficient outcomes.

CCI: How has your professional experiences (at Microsoft, HP and Google) influenced your academic career?
VG: These professional experiences helped me gain a deeper understanding of the "real world" constraints that arise when designing mechanisms. Google and Microsoft are two of the big players in online advertising auctions, so my work there focused on how to design such auctions, whereas my work at HP was related to the design of auctions for selling private data, while appropriately compensating the people to whom this data pertains. These positions provided me with the excitement of solving problems that are immediately useful within the corresponding company, but they also helped me appreciate the openness and long term vision that the academic world has to offer, which is what eventually led me to my current position.
CCI: What do you consider to be your biggest research accomplishment so far?
VG: One of the results that I am most proud of came during my PhD studies. I was working on a project aiming to tackle the problem of "fair division:" given a collection of divisible goods, one needs to design a mechanism that allocates them to a set of strategic agents in a fair way. But, how much each agent values each one of the goods is private information that only he or she knows. To ensure that the outcome is fair, the mechanism needs to ask the agents how much is their value for each good, but the agents can lie, if that ends up giving them a more valuable bundle of goods. Most known fair mechanisms could not ensure that the participants will not lie about their values, but we designed a mechanism that is surprisingly fair, while ensuring that no agent has any incentive to lie.
CCI: What courses are you teaching at Drexel right now and what do like most about them?
VG: During the fall semester, I taught an undergraduate class on "Data Structures and Algorithms," and this spring term I am teaching a graduate class on "Economics and Computation.” The undergraduate class was really exciting because I got to teach some of the most fundamental results in computer science, and I also got to meet some very smart undergraduate students. This term’s graduate course will be closely related to my research, and it will give me the chance to cover some of the results in my research field which I find most compelling. Also, a big part of this class will revolve around game theory, and the best way to teach game theory is to play "games" in class, i.e., ask the students to play the role of strategic agents facing different scenarios. It will be fun!

CCI: What’s your favorite thing about working in Philadelphia?
VG: Apart from the several good food options, I really enjoy how concentrated the city is around its center, and how the University City District combines two campuses [University of Pennsylvania and Drexel] into one. This makes both life and research much more pleasant and efficient, unlike some of the cities that I have lived in in the past.
CCI: Where do you see your research taking you in 10 years?
VG: Despite the surge of work on algorithmic game theory during the last 15 years, there are still several fundamental problems that the research community has not resolved. Some of my recent research has focused on a couple of these problems, but a deep understanding of the trade-offs and the techniques required will demand several more years of dedicated work. In 10 years from now, my goal is to have developed a robust and general theory that resolves these problems, while also training several new smart graduate and undergraduate students how to tackle problems using algorithms and game theory.

Media Contact