Quicklinks:

Computer Science Department’s 50th Anniversary Celebration

By Ted Pawlicki

Computer Science Department Timeline by Woody Wu

Special thanks to: Luke Auburn for his detailed editing suggestions and essential materials, Chen Ding for initiating the idea for the poster, Kristi Kongmany, George Ferguson, Shelley Zoeke, Derek Swanson, and Andrew McFarland for their helpful suggestions during the editing process. Click above to view.


Truly the highlight of UR’s 2024 Meliora weekend was the joyous celebration of the 50th Anniversary of the Department of Computer Science. The rise to greatness of UR’s Computer Science Department, from its humble beginnings in 1974, when computers were just being introduced for personal use, tracks the exciting era of the digital revolution during which computers went from being obscure tools for esoteric research to having the ubiquitous role they now play in our society at large.


The celebration of 50 years of computer science at the University of Rochester was as much forward-looking with excitement as it was a retrospective walk down memory lane. The event reviewed the numerous achievements that took place in the department. It also presented an exciting vision and purpose for the next 50 years and beyond.


This exciting whole day event on Friday, September 27th was dual themed. The morning session was dedicated to talks by some of the brightest luminaries affiliated with CS@UR.


The morning talks took place in Wegmans Hall, which has been the home of the Department of Computer Science since its construction was completed in 2017. Wegmans 1400, a 160-seat lecture hall was filled with notable personages sampled from the department’s past, present, and even its future, as evidenced by the many current students who attended. The wall outside of 1400 Wegmans Hall was adorned by a graphical timeline depicting some of the major highlights that took place during the 50 years of innovation and advancement in the department.


The afternoon session focused on fun, informal receptions where many old acquaintances were renewed and memories were exchanged. The atmosphere at these mixers was electric with excitement as conversations across areas and generations echoed the longstanding department culture of intellectual exchange and collaboration. The entire 50th Anniversary Celebration was memorialized with a commemorative T-Shirt designed by Brad Orego ’06 (BS).


Prof. Chen Ding, the current chair of UR’s Computer Science Department, served as Master of Ceremonies throughout the morning session, often sharing personal anecdotes about the speakers. Professor and Dean Wendi Heinzelman presented an introductory overview of the department’s history entitled “URCS - 50 Years of Memories” to tee off the gala event and set the tone for the day’s festivities.


The first talk was given by Daniel Sabbah ’74, ’82 (PhD). Dr. Sabbah graduated with a degree in mathematics just before the launch of the computer science department. He stayed at Rochester for graduate study in the new department and, after earning a PhD, went on to IBM. His talk, titled “‘Lessons’ in Computer Science” focused largely on the “Heart of Innovation.” It presented fascinating insights, gained during his career as an innovation leader, about the intricate dance between technical advances and societal/market forces. In keeping with the ever forward-looking culture of the event, Dr. Sabbah has made a $2 million contribution to establish the Distinguished Professorship in Computer Science. This generous gift, which represents the Computer Science Department’s first departmental endowed chair, will certainly help vault the department into an ever brighter future.

Distinguished Speakers

Wendi Heinzelman (Dean), Daniel Sabbah, Amanda Stent, Chris Stewart, Michael Scott, Chen Ding (Chair)


The second talk of the morning session was given by Dr. Amanda Stent ’01 (PhD) who is currently Head of Artificial Intelligence Strategy at Bloomberg LP. Her talk, entitled “Computing in Context” began with a number of observations on some of the great features of the URCS culture. In keeping with computer science norms of presentation, and much to the delight of her audience, Dr. Stent’s talk was organized as a two level trinary tree (three major parts each with three sub-parts). In addition to expressing her thanks and appreciation to the Computer Science Department, she made several insightful observations on computer science and artificial intelligence. As warm and hopeful as Dr. Stent’s talk was, she did not shy away from controversy. Her talk raised a number of challenging questions for the entire field of computer science. These included questions vital to public policy in general and the field of computer science in particular. They included topics ranging from “why do so few people want to go into computing?” to the critical societal challenge of fostering responsible artificial intelligence research. She concluded her presentation with lessons learned for computing careers.


The second half of the morning session was kicked off by Dr. Christopher Stewart ’09 (PhD), now a professor of computer science and engineering at Ohio State University. He warmly expressed his appreciation for URCS for both its technical excellence and the personal impact it made on him as a scholar. His talk, entitled “Unlocking Planetary Digital Agriculture Transformation” gave a fascinating overview of data driven methods used to realize sustainable food security worldwide. Dr. Stewart presented his group’s ambitious 3-dimensional research agenda aimed at solving many of the looming problems found in global food insecurity. He highlighted projects ranging from diagnosing crop heath with artificial intelligence to performance modeling of swarming Unmanned Aerial Vehicles (UAVs). His work presented a challenging invitation to new researchers in this burgeoning area that combines challenging problems across geo-spatial, Artificial Intelligence complexity, and the systems stack.


The final talk of the morning session was presented by Dr. Michael Scott, the Arthur Gould Yates Professor of Engineering and Past Chair of the Computer Science Department at the University of Rochester. His talk, appropriately titled “URCS @ 50” reviewed five decades of glory in 30 minutes. Dr. Scott joined the Computer Science department at the University of Rochester in 1985 after earning his PhD from the University of Wisconsin-Madison. With his 39 years of direct personal observation, Dr. Scott was uniquely qualified to review the department’s history. He was quick to point out the multi-dimensional aspects of the Computer Science department’s history, offering perspective through the lens of buildings, people, hardware, research, academics and administration. Always looking forward, he concluded with a snapshot of where the department is today and where the trends and constraints of today will take the department in the future. Dr. Scott is a well known champion for the collaborative consensus based culture that is a distinctive feature of the Computer Science Department at the University of Rochester. He often opines: “Computer science is an interdisciplinary field, but I think it’s more the case here than almost anywhere.” Interested readers can view the recordings of the morning talks on the Computer Science department’s website.


After lunch, the 50th Anniversary celebration continued with a “Meet Future Leaders” event in the 2nd floor Atrium of Wegmans Hall, which has become a popular “hang out” location for undergraduates in Computer Science since the move to Wegmans Hall. This inspiring event brought together former and future leaders in computer science that lifted the aspirations, hopes and dreams of younger scholars while assuring the more senior graybeards that the future is indeed in good hands.


The final event of the 50th Anniversary celebration was a general reception for all the participants in the luxuriant Hawkins-Carlson Room in Rush Rhees Library. Participants enjoyed themselves over good food and lively conversation. This joyous gathering was a testimony to the significant impact of the past achievements of and the bright future ahead for UR’s Department of Computer Science.

50th Anniversary Reception, Hawkins-Carlson Room. Photo by Sreepathi Pai

Featured Article

Speeding into Trouble:

The Side-Channel Risks of High-Performance Processors

By Yanan Guo

Processor Side Channels 101


When you think of leaking secrets stored on a computer, you might immediately picture attackers exploiting software vulnerabilities to gain access to sensitive data. But that’s not the only way. A less obvious yet equally dangerous approach is processor side-channel attacks. Instead of accessing the secret data, these attacks extract the secret by exploiting the subtle clues left behind in the hardware during data processing, such as timing variations, cache behavior, or power consumption. Imagine someone listening to the clicks of a lock to figure out the combination—that is how side channels operate. For example, just by monitoring the execution time of an encryption algorithm running on the CPU, an attacker can infer the private key used in the encryption. In this article, we will dive into how side channels work, why they have become a serious threat, and the race to uncover and counter these stealthy attacks.


1. Why are there side channels in the processors?


The root cause of side channels is two-fold. First, modern processor designs prioritize performance and efficiency, often treating security as an afterthought. Modern processors use many techniques to optimize performance and efficiency, such as overlapping tasks, predicting the next steps in computation, and caching frequently used data for faster access. These innovations are key to the speed and power we enjoy today. However, they also create subtle variations in the processor’s behavior depending on the specific data being processed. These variations, when observed by an attacker, can leak secret information. For example, during a password entry process, the cache access pattern is influenced by the specific key being pressed: different keys result in different memory addresses being accessed, which in turn occupy different cache locations. An attacker who can monitor changes in the processor’s cache can infer which locations were used and, as a result, deduce the keys that were pressed.


The second key factor driving side-channel vulnerabilities is resource sharing. In modern computing, especially in cloud environments, multiple processes or virtual machines often share the same hardware platform, such as a CPU. While this boosts efficiency, it also creates a dangerous opening for attackers. By running malicious code on the same system as the target, an attacker can monitor shared hardware components and detect changes in their state. For instance, to discover which cache locations are being used by a password entry process, the attacker can observe which cache locations their own data is being evicted from due to cache conflicts, allowing them to infer sensitive information like the entered password without directly accessing the data.


2. Why are side channels a serious concern?


It has been shown that side channels have far-reaching consequences, impacting a wide range of applications and leaking highly sensitive information. For example, in cryptographic systems, side-channel attacks can be used to extract encryption keys by observing power usage or cache access patterns during cryptographic operations. This compromises the security of data that relies on encryption, such as secure communications and financial transactions. In password entry processes, attackers can infer keystrokes and capture passwords. In AI applications, side channels can be exploited to leak various types of information, including DNN architecture and weights, threatening the confidentiality of AI models. Web browsers are not immune either—researchers have demonstrated that side channels can be used to recover sensitive information displayed on active web pages, such as user credentials or other personal data, posing a significant risk to privacy and security.


3. Is your laptop vulnerable to side channels?


Likely, yes.


Spectre: The Big Heist on Modern CPUs


1. What is Spectre? How does it work?


Spectre is one of the most well-known side-channel attacks on modern CPUs. It exploits a common feature in CPUs called speculative execution: CPUs make speculative guesses and execute instructions ahead of time, based on the assumption that those guesses will be correct. If the guesses are wrong, the CPU rolls back the changes—but crucially, some side effects of these speculative operations (such as changes to the cache) are not fully rolled back. This is where Spectre comes into play.


In Spectre, the attacker manipulates the speculative execution process by tricking the CPU into executing instructions that would not normally be allowed to run, and accessing data that could not be normally accessed. While these speculative instructions will eventually be squashed, they leave traces in the hardware. By carefully measuring these traces, an attacker can infer the values of sensitive data that was accessed speculatively, such as passwords or cryptographic keys.


2. The history and current status of Spectre.


The Spectre vulnerability was discovered independently by two groups of researchers in 2017 and publicly disclosed in January 2018. Its discovery sent shockwaves through the tech industry: it revealed that almost all modern CPUs are vulnerable to some form of side-channel attack due to speculative execution, a core feature in nearly every high-performance CPU produced over the last decade. Spectre is often considered a more significant concern than earlier side-channel attacks. This is because, rather than relying on the victim program to access the sensitive data, Spectre allows attackers to manipulate the CPU itself, tricking it into accessing the data that the attackers are interested in.


In response to Spectre, hardware and software vendors have developed a range of mitigation strategies, many of which are now available in modern computing systems. Examples include Retpoline, a software-level technique designed by Google, and IBRS, STIBP, and IBPB, which are hardware-level designs from Intel. Despite these efforts, Spectre remains a significant security concern today: although mitigations have been deployed (at a high cost to performance), researchers continue to discover new variants of the attack that bypass one or more existing defenses.


Current and Future Research on Side Channels


1. Uncovering side channels in modern processors.


Currently, many researchers focus on reverse engineering processor designs (CPUs, GPUs, and beyond) and uncovering previously unknown side channels. As these processors become more and more complex and powerful, they often introduce new security vulnerabilities alongside their performance advancements. When researchers identify a new side channel, they typically report it to the vendor before publicly disclosing the details. This allows vendors to develop and deploy patches to address the issue. In some cases, vendors may request an embargo period before the vulnerability is disclosed, giving them additional time to develop and implement fixes. This collaborative process prevents attackers from taking advantage of newly discovered weaknesses.


2. Developing defenses at software and hardware levels.


Alongside efforts to uncover new vulnerabilities, another major focus of research is developing defenses against side channels. At the software level, techniques such as constant-time programming have been introduced to mitigate side channels by ensuring that code execution time is independent of sensitive data. On the hardware side, computer architects are designing secure processor architectures that are more resilient to side-channel attacks. For example, one approach involves dynamically partitioning hardware resources between users, preventing attackers from observing hardware state changes caused by the victim’s data. Hardware solutions typically introduce less performance overhead compared to software-based defenses; however, they also take longer to deploy.


3. The cat-and-mouse game continues, but new hopes emerge.


Despite the progress made in developing defenses, we are still engaged in an ongoing cat-and-mouse game. As new defenses are introduced, offensive security researchers often identify flaws and develop new attack variants that can bypass these protections. However, there are promising developments on the horizon. One such advance is formal verification, a technique that uses mathematical proofs to verify the correctness of a system. In the context of side channels, formal verification can help ensure that critical hardware features, like speculative execution and memory access, are free from potential leaks by design. Although this field is still in its early stages, it holds the promise of a future where side-channel vulnerabilities are significantly reduced and even eliminated. This shift could move us beyond the reactive cycle of constant patching toward a more proactive and robust security model.


Conclusion


In a world where performance and efficiency drive modern processors, side channels remain a persistent threat, turning hardware optimizations into vulnerabilities. As we continue to push the boundaries of what processors can do, it becomes increasingly important to understand and mitigate side channels to ensure a secure digital environment. Over the past decade, researchers have dedicated significant effort to this field, developing numerous defense mechanisms. Although these defenses are not yet perfect, there is hope that, with continued research and innovation, lasting solutions to these challenging problems will be found.


Len Schubert Retires

By Dan Gildea

In June, the department celebrated the retirement of Len Schubert, one of the pillars of the artificial intelligence group at Rochester since 1988. Len’s work investigated many issues in machine reasoning and natural language processing, including methods for extracting knowledge from language and guiding the interpretation of language with knowledge. Len has made major contributions in the famous “frame problem,” which considers how to reason about which properties change and which can be assumed to stay the same from one moment to the next. Len also developed an influential theoretical framework for representing the agents’ beliefs in logic and reasoning about them. More recently, Len has worked in the area of extracting commonsense knowledge about the world from large amounts of text - a problem made more difficult by the fact that people tend not to explicitly state background knowledge that can be assumed on the part of the reader. Len has collaborated with other members of Rochester’s faculty to develop a number of interactive conversational systems. These include the groundbreaking TRAINS and TRIPS systems for computer-assisted planning in the 1990s, and more recently the LISSA system for individuals with ASD and the SOPHIE system for patient interactions in healthcare. Throughout these projects and many others, Len’s work was marked by a focus on the long-term goal of systems capable of autonomous, intelligent reasoning in the service of humanity.

Len’s dissertation, in aerospace engineering at the University of Toronto, investigated diffraction of sound and developed new methods for numerically solving general nonlinear systems of equations. Len made a leap from the continuous to the discrete as a postdoc in computer science at Johns Hopkins University, where he worked on recursion theory, randomization, and issues of minimum-length programs. Len then spent the period from 1973 to 1988 as an assistant, associate, and then full professor of computer science at the University of Alberta. Looking for a smaller, more collegial environment, Len moved to the University of Rochester in 1988. He worked with James Allen on many projects as part of the Conversational Interaction and Spoken Dialogue Research Group. At the University of Rochester, Len advised 19 PhD dissertations, and many of his former students are now prominent researchers in their own right. He contributed to the department behind the scenes as well, in particular in graduate admissions committee, and he was an advocate for underrepresented groups in the field.

While we marked Len’s official retirement with a celebration in the department in June, we do expect and hope that he will continue to be involved in his role as professor emeritus.


Congratulations, Len.

Promotions and Departures

Featured Graduate Alumnus

Michael C. Chavrimootoo

By Lane A. Hemaspaandra

The year covered by this issue included, during the 2024 Meliora Weekend, URCS’s 50th Anniversary Celebration. That exciting milestone in the department’s history is reported on in this issue in the article by Ted Pawlicki. It was a wonderful day, attended by many alums and celebrating a remarkable half century of research and education. Among the morning sessions talks were three by tremendously accomplished URCS Ph.D. graduates - Daniel Sabbah, Amanda Stent, and Christopher Steward - and indeed the department over the past 50 years has produced a huge number of wonderful undergraduate and graduate alums who have gone on to advance CS research and education through in their positions in academia, industry, and government research labs. It is a history to be very proud of. However, it is not just the department’s past that is filled with remarkable students. The department continues to attract and educate wonderful students at both the undergraduate and graduate levels. To celebrate the present, this issue’s faculty highlight is on the work of newly hired Professor Yanan Guo, and this issue’s grad alum and undergraduate alum interviews diverge from Multicast’s norm of interviewing alums who are many years beyond their UR education, and instead each highlights one of the many truly extraordinary students UR is currently educating and graduating. In particular, the undergrad alum interview is with May 2024 URCS B.S. graduate Anisha Bhattacharya of Morgan Stanley and the grad alum interview is with May 2024 URCS Ph.D. graduate Michael Chavrimootoo of Denison University. Here’s to 50 more wonderful years of contributions to science, education, business, and society by URCS alums and faculty!


Michael C. Chavrimootoo received his B.S. in Computer Science and his B.A. in Political Science from UR in 2020, his M.S. in Computer Science from UR in 2021, and his Ph.D. in Computer Science from UR in 2024. While a graduate student, Michael spent his summers teaching at UR and RIT. During his graduate studies he won UR’s Edward Peck Curtis Award for Excellence in Teaching by a Graduate Student, UR’s Graduate Research Day Best Research Talk Award, and UR’s Outstanding Dissertation Award, and also was recognized three times by RIT’s Teaching Gold Standard. In the summer of 2024, he became a tenure-track assistant professor of computer science at Denison University. Michael’s research interests span computational social choice, algorithmic game theory, and computational complexity theory.


Michael was interviewed for Multicast by URCS professor Lane A. Hemaspaandra, who was Michael’s Ph.D. advisor.


Lane: It’s great to be chatting with you, Michael! To set the scene for the coming questions: This issue covers URCS’s 50th anniversary. The celebration we had for that featured talks from some tremendous URCS Ph.D. alums of the past 50 years. However, as you’ll know from your time at the university, UR has not only a 50-year past of terrific students, but also a terrific present, with students at both the undergraduate and graduate levels often not merely taking courses but also doing truly important research. To stress the excitement of not just the past but of the present, the undergrad and grad alum interviews this issue are with very bright students who just now graduated. And the grad alum is you. No pressure! :-)


The introduction to this article mentions the amazing hat trick of awards you won at UR: for research, for research presentation, and for teaching. Let’s start by taking about each of those three aspects of your time here.


Turning first to research, you won UR’s Outstanding Dissertation Award. Could you tell us about the computational social choice part of your thesis research: What did it do, and why is that important?


Michael: My contribution to computational social choice revolves around the study of electoral control, which is a type of electoral attack involving an entity that has the ability to alter the structure of that election by adding/deleting/partitioning voters or candidates. The work on (electoral) control started back in the 90s, with researchers often looking at the complexity of making a candidate win using those types of attack. Such research on control has been active ever since, with even more types of control attacks being considered, as well as a wide range of voting rules. This is a tremendously important line of work, as it models key and contemporary issues in electoral integrity, such as voter suppression, gerrymandering, and vote splitting, among others.


My thesis focused on looking at so-called separations and collapses among these control attacks. In some sense, we looked at whether pairs of control attacks could be, for all intents and purposes, equivalent under various notions. What does mean? Let’s look at a simple example.


Let’s say that we have two candidates in our election, say - I’ve chosen the names arbitrarily - Lane and Michael, and 11 voters. We’ll use a simple scheme (aka voting rule) to aggregate the preferences of the voters: Each voter selects a candidate, giving that candidate one point, and we’ll say whichever candidate(s) get the most points are the winner(s). (This “scheme” is also known as plurality voting.) So if six of our 11 voters pick Lane, and five pick Michael, Lane gets six points, and Michael gets five points, making Lane a (and indeed, the) winner. Now, suppose one were to seek to make Michael win by deleting voters. A successful action here would be to delete two voters who picked Lane, as Lane would now have four points, and Michael would win with five points. This models highly targeted voter suppression, and finding that action isn’t hard in general when the underlying voting rule is plurality: there is in fact a natural polynomial-time algorithm for control by voter deletion. However, if one wants to guard against such an attack, one had better pick the voting rule in a way that likely makes computing the attack computationally difficult (and most papers on this seek to make attacks be NP-hard).



This work on separations and collapses is important for several reasons. If two control attacks are in effect the same, that is they collapse, it does not make sense to study them separately, even though as we found in our work, there were many collapsing control attacks that were studied separately in the literature over the years, since no one had realized that they collapsed!


Across the four categories we looked at, we showed that there are 15 additional pairs (of control attacks) that collapse---in addition to the known 28 pairs that collapse---and we proved that there aren’t any other pairs (of control attacks) that collapse across those four categories. We even took it a few steps further by refining our separation results to a “finer granularity,” and we also proved that some of our collapses followed from simple axiomatic properties, which technically gives us an infinite number of collapses among an infinite number of categories.


Perhaps most surprising to me was that we proved results that showed how to transform a successful attack that was about partitioning voters into a successful attack about partitioning candidates, which one would think should not have meaningful relationships, given that we’re partitioning different components of the election! Partitioning is a particularly interesting topic, since it models gerrymandering in some sense, which is a topical issue here is Ohio!


Wow, I meant to be brief here, so I guess there are a lot more interesting things to say than I realized. :-)


Lane: Wonderful! And now can you tell us about the core-complexity part of your thesis, which explores a long-open issue about different levels of nondeterministic ambiguity? It is pretty technical stuff, but can you informally convey in flavor what did it do, and why that is interesting?


Michael: Ah yes, absolutely! Let me try to give an example that captures the essence of the problem.


Let me pick another arbitrary name here. Let’s say Audrey is a maze enthusiast, and loves finding her way through “mazes.” Let’s assume Audrey really cares about knowing whether there is a way out of the maze. Recently, Audrey has been enthralled by a specific class of mazes: those mazes have one entrance, one exit, and at most one path (possibly zero) from the entrance to the exit. We have to make a few assumptions here, so let’s just pretend that for some mysterious reason, perhaps due to the properties of the maze (which in truth are not explicitly written but are exponential-sized, implicit structures induced from an input via nondeterministic computation trees and some additional construction steps), she is not able to apply any of the known, convenient path-exploration algorithms. Let’s use the notation Mazes[1] to denote this class of mazes, and let’s analogously define Mazes[2], Mazes[3], etc. Now, Audrey is convinced that there is a procedural way of figuring out whether such a path exists in a maze from Maze[1] without taking too much time. Now, the interesting question is the following one: If such a procedure exists, does that imply the existence of a “not taking too much time” procedure for Maze[2]? And what about for about Maze[3], and so on?


It turns out we have an answer to that question. Finding a path in the kind of maze we are speaking of is a type of NP problem, since one can easily verify if a given sequence of directions forms a path from the entrance to the exit. Leslie Valiant defined the complexity class UP, which is the class of languages accepted by NP machines having at most one accepting path. So, in some sense, finding a path in a Maze[1] maze is a UP problem. The “at most one” requirement of UP can of course be generalized using some type of reasonable function over the natural numbers, to get other classes with requirements (i.e., ambiguity bounds) of the form “at most 3” or “at most n squared, where n is the input’s size.” And finding a procedure that doesn’t take too long is effectively asking whether there is a polynomial-time algorithm to tell us if there is a path from the entrance to the exit. Roughly speaking, Osamu Watanabe proved that for each natural number k greater or equal to 1, there is such an efficient procedure for Maze[1], if and only if there is such an efficient procedure for Maze[k]; he in one fell swoop linked the fates of an infinite collection of problems and their relation to deterministic polynomial time (aka P). It had been open for decades as to whether we could improve on such types of linkages, and my coauthors and I gave two results that provide new ways of creating “linked fates.” I won’t burden your readers with the details, but I’ll say that one of those linkages felt like a white whale to me. I spent months convinced that the linkage had to be true, but I had no proof; I tried various approaches that all failed. Then one day, during one of my three or four flights from Mauritius to Rochester, I suddenly had a realization that helped me figure out the right approach! I’m sure there’s a lesson somewhere in there...


Anyway, the work on ambiguity-bounded computation is truly fascinating, and we’re still actively working on more results there!


Lane: Way-cool! And can you tell us about the game-complexity/“defying gravity” part of your thesis?


Michael: I sort of fell into that area by accident when looking at a specific paper that caught my attention; I wasn’t familiar with the study of the complexity of video games, so I was quickly intrigued. One type of game commonly studied is those where “winning” typically involves moving blocks or objects to achieve a goal, under the rules of the game. And some games include a gravity component that kicks in and makes your blocks fall unless they’re appropriately supported, and there is no direct action in the game to undo that (so, no jump feature). In such games, the number of possible configurations is restricted, since some are not realizable. Nonetheless, it seems like the complexity of such games is not any easier than their without-gravity-and-fully-reversible counterparts, which have been, and still are, heavily studied. The most popular approach to studying the complexity of fully reversible games uses graph-theoretic methods, and it is extremely attractive as it allows one to give hardness results using only two gadgets. That approach involves the use of NCL (nondeterministic constraint logic) graphs, and unfortunately its application to games with gravity isn’t always simple. I realized at some point that by using additional methods from graph theory, one could add structure into the NCL graphs, and that allowed one to “abstract away” the gravity part of the game enough to allow one to reuse knowledge from the study of fully reversible games. You can see from the images that an NCL graph can be very simple, but in contrast, the gadgets look more complex.

Interestingly, that was not the only challenge that had to be overcome, as adding structure to NCL graphs has a nasty side effect that makes direct reductions far more complicated. It creates a blowup in the number of gadgets needed, going from only two up to 32. However, it turns out that if one looks at the bigger picture, there are hidden underlying links that one can exploit to almost completely get rid of that nasty side effect by getting the number of gadgets needed down to three! That work is independent of the game being analyzed, so it has the potential to be applied to other games with irreversible gravity! This opens up avenues to other interesting problems surrounding both the specific game I studied, namely the Hanano Puzzle, and other games with irreversible gravity.


Lane: So we’ve now seen that you wrote a triple-threat thesis: elections, ambiguity, and games; wow!


I said I’d go through the areas of your three awards, and you just covered the area of research. But as to research presentation and the best research talk award that you won for that, I see that your answers above, both via your text and the poster and image you sent, already make pretty clear why you won UR’s best research talk award! So let us right move on to the area of teaching, something I know you are passionate about doing well, and in which you won UR’s prestigious Edward Peck Curtis Award for Excellence in Teaching by a Graduate Student. Regarding the latter, there are three types of teaching that a student can win the award for: teaching, TA-ing, and mentoring research students. Returning to our theme of threes, I believe you were a rare triple-threat for the award as you excelled in all three of those areas: As a course teacher, you had remarkable student ratings at UR, and thrice won RIT’s Teaching Gold Standard. Can you tell us what approach you take that makes your teaching so successful? As a TA - and I’ve had the treat of having you as a TA both when you were an undergrad and when you were a grad - you were really valued by and supportive of the students in the courses. Can you give any advice to readers - such as our current UG students - as to how you approached TA-ing so successfully? And finally, as to mentoring research students, you’ve worked closely with about a dozen and a half URCS undergraduates, coauthoring technical reports, conference papers, or journal papers with most of those. That is amazing. What is your approach to helping undergraduates find their voice as researchers? (And yes, that is three questions!)


Michael: Aha, a triple-threat question!


I’m not sure if I have good advice, but I’ll take a stab at it. At the center of it all, I approach my teaching (and TA-ing) in a service-oriented way; that does mean that it takes a lot of time to provide adequate help, which takes on various forms, such as explaining the material, encouraging students, supporting students through their efforts, following-up on what they do and their interests, challenging students who grow complacent, inspiring students to see the material through a different angle, etc. Moreover, one thing that has made a huge difference in the quality of my teaching is realizing that the way I learn and process information is not the way my students learn and process information, so figuring out ways to facilitate their learning requires a bit of creativity sometimes. If I’ve learned anything over the last two months, it is that there are so many ways in which I can improve, and I’m still learning how.


As to research, I found that the most successful students were deeply interested in the research area, engaged critically with the work, and put in consistent effort. This feels like a no-brainer, but I think it’s easy to forget to do these three things, especially the “consistent effort” part. Nonetheless, adopting those habits and asking questions (lots of questions) is a great way to find your way in a sea of information where a lot of work has already happened. So, when working with students, I try to encourage that a lot, and in a way my teaching habits have helped me mentor those students.


Lane: You just now covered a bit about how you have mentored students while at URCS. Let’s turn to how you were mentored. In particular, in addition to three computational social choice researchers (A. Kahng, E. Hemaspaandra, J. Rothe) and a mathematician (S. Lubkin), you had on your committee URCS’s wonderful Len Schubert, who after many decades at UR retired this summer Can you share with us any lessons you learned from Len that helped you during your time at UR?


Michael: Early in the fall of 2020, you said something to me along the lines of, “Len embodies what it means to be a scholar, and we should all strive to be like him.” Or maybe that’s not what you said, but it is what I took away. I took a course with Len that same semester and quickly understood what you meant. The first thing that jumps to mind is just how careful and critical he was. Throughout the course, there was always a purpose to everything we did, and he was just so methodical about it. He would ask simple questions that required us to develop a deep understanding of the subject matter to be able to respond. That became even more obvious when he joined my thesis committee. Every time we met, Len was always so humble, often prefacing his questions with something along the lines of “This may not be a smart question,” and yet his comments and questions always threw me off-guard, because they expressed incredibly deep thoughts. I learned not just about intellectual humility from him, but also about the importance of spending considerable amounts of time thinking about the fundamental components of my research.


Lane: Yes, Len has always been a shining model - as a person, an educator, and a researcher - to all of us, faculty and students alike. His wise, kind, insightful presence in the department will be tremendously missed.


Returning to the theme of threes, your undergraduate, masters, and doctoral degrees are from the University of Rochester, for a total of eight years in Rochester. In case some of our current students are reading this, and as tips for future Meliora Weekend attendees, are there some not-to-be-missed gems of UR, or of Rochester itself, that you’d like to commend to our readers? And, specifically for our current (and future) students, do you have any advice - whether on what to do or what not to do - on how to be a successful student at URCS?


Michael: Now, I cannot give away all my favorite gems, but here are a few highlights. At UR, make sure to attend the spectacular shows at Eastman, including the student performances, and to explore the various parts of the campus. There is a list, though it may be a bit outdated, of 101 things to do before graduating on the UR website; that would be a good place to start.

As to Rochester, there is so much to it. There’s a history tour of downtown, an art walk downtown, a tour of Mt. Hope Cemetery and the stories it hides, boat tours on the Erie Canal, amazing breweries and coffee shops, and a stellar food scene in the Henrietta region, especially if you’re looking for non-American and non-European food! There are some really fun bike trails in the area too; I highly recommend biking from campus to Lake Ontario on a cool fall day. Oh gosh, I could say a lot more about gems in Rochester and the surrounding area, and I do encourage any current student to go out and explore the area; it makes for a significantly richer experience at UR! If you’re looking for recommendations, feel free to contact me. :-)


As to being successful at URCS, while there are many ways of doing that, it’s important to develop relationships within the department - and of course outside the department - both with professors and peers, as those are the people who will be instrumental to your success in one way or another. Forging those relationships also allows you to have more opportunities to collaborate and discuss a diversity of ideas. Also don’t overcommit. It’s a trap that a lot of UR students fall into (the past version of me included), and once they’re overcommitted, they struggle to apply themselves and excel in some truly important ways.


Lane: Goodness, you just added a lot of things to my own to-do list in Rochester; thank you!


Having been a URCS student for eight years, and a URCS instructor-of-record for three summers, do you have any favorite anecdotes about your time at the department as a student or as a teacher?


Michael: A turning point in my time at URCS came while I was an undergraduate, when I spent a summer working for Chen Ding; that was the point where I decided I wanted to become a CS professor. During that summer, some of the other students in the group and I had developed a habit of playing foosball, almost religiously, during our lunch breaks. It was a great time for us to socialize and get to meet people, especially the grad students at the time. Those breaks also helped us develop a sense of community and foster a spirit of collaboration during the summer. And we were all very competitive, so by the time our break was over, we had this surge of energy to tackle our research problems. I’m still in touch with some of these folks even though it’s been a few years, and I think back to those moments very fondly.


Lane: So far, we have focused on your time at the University of Rochester. However, you just started as a tenure-track CS faculty member at Denison University. You’re only two months into your first semester there, but can you share with us how you’re liking Denison so far, what the differences are between teaching there and teaching here?


Michael: Denison has been better to me than I ever expected, and it’s been a great transition. Overall, the community is incredibly supportive and dedicated to providing an excellent and well-rounded education to students, and in getting high-quality research done. Perhaps my biggest lesson since joining Denison has been learning that there is a lot of room for me to grow as an educator, both inside and outside the classroom, and I’m learning more about that every day. As to the differences between teaching at Denison versus URCS, the biggest one has been transitioning to a model that encourages far more opportunities for active learning. For comparison, I think my only URCS classes that incorporated a lot of active learning were with you Lane, but here, it’s all of them! I’m finding that I have to relearn how to teach certain topics, even though I’ve taught them a lot at UR or at RIT. However, more active learning does mean that you can’t cover quite as much material, but I think the active learning component allows for a deeper understanding of the material covered.


Lane: I very much agree (and indeed we, RIT, and UR’s Warner School of Education have an in-progress grant on evaluating whether active, model-based learning is effective in helping students better reach theoretical computer science literacy).


The following is a bit of an unfair question, since you’re only two months into your time as a professor, but nonetheless, what as an educator and what as a researcher would you most like to accomplish during your academic career?


Michael: It’s a good question, and I’ve been pondering that a lot recently, especially since tech is having increasingly more influence on our everyday lives. As an educator, I’m of course thrilled when students do well in their classes and ace material they found hard, but I’m also interested in seeing how we cultivate certain values within students. For example, how do we teach students to think about the ethical and moral implications of their work? How do we teach students to take these strong theoretical foundations that we’re giving them to effect change in society or to solve problems in the world, whether that be in industry or in academia? Another reason I love being at Denison is because I’m surrounded by folks with similar goals and intentions! As a researcher, I’m interested in developing models that are rooted in strong theoretical foundations to study contemporary “problems” in elections. For example, what are some natural and convincing (to the public) methods to draw out districts? Can we come up with natural and reasonable properties for such methods? And what computational properties do they have, e.g., what is the complexity of carrying out certain attacks given a redistricting method?


Lane: What exciting and important ideas and directions!


I notice you didn’t include resolving P versus NP. Nonetheless, here are some lightning-round questions: Do you think P versus NP will be resolved via equality, or via inequality, being proven? What is your best guess for when that resolution will be obtained? If you are predicting P is not equal to NP, how much time do you think will pass between that being resolved and the resolution of the issue of whether the polynomial-time hierarchy collapses (though perhaps to a level higher than NP)?


Michael: I only recently read Lance Fortnow’s “The Golden Ticket” book that talks about P versus NP in a nontechnical manner, and although he paints a picture of a quite beautiful world if P=NP, I think the problem will be resolved by proving that P differs from NP. As to when, I would say within the next sixty years, but only because I would like to see it happen during my lifetime, and because I hope to be so lucky as to live for that long! For the polynomial-time hierarchy, I’ll venture a guess that it’ll be resolved within 20 years of the resolution of P versus NP, but that may be optimistic.


Lane: It would be hard to imagine that you had non-work time during your eight undergrad/grad years here, given how much you accomplished academically. Yet, in fact, you were extremely active in doing service to the community, for example helping raise money for and yourself personally participating in the rebuilding of a school in Nepal, being a project coordinator in your home country of Mauritius for Experiential Learning Initiative Africa, serving for seven years as a care provider and trainer for Students Offering Support, and much, much more. It is a really remarkable record of caring service to community and society - one not of mere words, but of actions. However, as enjoyable and rewarding as your educational/mentoring/scholarly accomplishments and the just-mentioned deep commitment to service must be, are there perhaps some additional enjoyable and rewarding things, such as hobbies and activities, that round out your schedule?


Michael: I discovered while at UR that I had a passion for cooking, and I spent a considerable amount of time expanding my range of knowledge on recipes, ingredients, techniques, etc. In fact, for three semesters, and maybe a summer or two, I was one of two people who volunteered to cook all the meals at the Newman Community’s events, of which there were about 6 to 8 a month! It’s a fun hobby at the end of the day: I get to have a (usually) delicious meal! But apparently it’s not a healthy hobby, so I also picked up running and biking over the last few years. I love being outside, so I think it fulfills several needs; talk about being efficient! I think that the nature of our work easily allows us to spend endless hours on our research, so working in those little breaks every so often actually helps with my productivity!


Lane: Thank you so very much for doing this interview, Michael! Continue to - in the kitchen, on the track, in the classroom, and in your research lab - do great things!

Michael and his partner Audrey, after running the Wilmington (NC) Half-Marathon in 2023.

Featured Undergraduate Alumnus

Anisha Bhattacharya

By Monika Polak

Anisha Bhattacharya earned her BS in Computer Science in May 2024, graduating with distinction. She recently began her role as a Technical Analyst at Morgan Stanley in the New York City Metropolitan Area.

Anisha was interviewed for Multicast by URCS professor Monika Polak, who had the pleasure of knowing her through the Cryptography and Post-Quantum Cryptography courses, where Anisha performed exceptionally well.


Monika: What are you currently doing in your professional career? Can you describe your current role and responsibilities in your job?


Anisha: I work as a Technical Analyst at Morgan Stanley, where the company provides wealth management services and technologies. They have an extensive training program that lasts four months. What I really appreciate is that they bring in instructors from all over the world. The company’s system is highly advanced, involving many technologies that require thorough preparation and understanding before you can start working. That’s why they have such a comprehensive training program.


Monika: Has your degree in computer science helped shape your path?


Anisha: Yes, I will be working as an application development programmer. Many of the classes I took in the CS program, especially Java programming, Data Structures, and Algorithms, have prepared me well for this role. These courses gave me a solid understanding of fundamental computer science concepts. Another very useful course was web design.


Monika: Which skills or knowledge gained from your computer science degree do you use most in your current role?


Anisha: Definitely coding in Java, since banks rely heavily on Java for their systems.


Monika: Were there any courses or projects during your studies that had a significant impact on your career?


Anisha: One of the most impactful projects was from my Data Structures and Algorithms course (CS 172), taught by Professor Ted Pawlicki. We had to implement Dijkstra’s algorithm to visually simulate traffic flow, and I received a lot of questions about this project during my interview.


Monika: Can you share an example of other projects or product you’ve worked on that you’re particularly proud of?


Anisha: When I was a Research Intern at the University of Rochester, I developed Deep Learning strategies to analyze trending conspiracy theories on Twitter. In my web development class, we created a buying and selling platform for UofR students, where they could sell items after moving out of the dorms. Although we never launched the platform, we were really proud of the outcome of the project.


Monika: How did the computer science department contribute to your success?


Anisha: The courses I took in the computer science department were critical for developing the skills I need in my current job. I felt well-prepared for my interview questions, and I also met a lot of great friends who supported me. One of the most valuable experiences was being a workshop leader. It pushed me to refresh my knowledge weekly to help other students understand key concepts. This role gave me a deeper understanding of the material.


Monika: What advice would you give to current or prospective computer science students?


Anisha: It’s helpful to observe how others apply for jobs to understand the process better. I gained a lot from visiting the tutoring center CETL —not only did I improve my skills, but I also connected with others. I’d also recommend getting involved in teaching positions, as it’s a great opportunity to master your skills while helping others.


Monika: If you could go back, is there anything you would have done differently during your time as a student?


Anisha: I wish I had started connecting with older peers and participating inside projects like hackathons earlier. Those experiences are really valuable for skill-building.


Monika: Where do you see yourself in five or ten years within the field of computer science?


Anisha: I would prefer to work in a place that creates new technologies, rather than just making changes to existing ones. In five to ten years, I could see myself at a big tech company or a startup—maybe even one I start myself. I want to do more than just minor updates to current systems.

2024 Commencement

University of Rochester photo/ Lauren Petracca

Undergraduate Awards:


Award for Outstanding Teaching Assistant: Justin Tsai

Entrepreneurship Award: Thanothii Ganesh


Excellence in Undergraduate Research: Sangwu Lee, Pinxin Liu, Yumeng He, Alexander Martin


UR Women & Minorities in Computing Leadership Award: Michelle Reingold


Most Valuable Programmer: Zeyu Nie


Outstanding Senior: Angie Pham and Sifeng Chen

Recent PhD Conferrals

2024 Honors Research

We had six honors students in the Undergraduate Class of 2024.


Steven Oufan Hai: Fingerprinting VPN Traffic: An Evaluation of Website Fingerprinting Attacks on Modern Virtual Private Network Applications


Junfei Liu: Causal Dataset Discovery with Large Language Models


Pinxin Liu: An Empirical Analysis on Large Language Models in Debate Evaluation


Fangyu Luo: Exploring Hyperparameter Tuning: A Survey and Experimental Framework Utilizing Multi-Objective Multi-Armed Bandits


Alexander Martin: Human-Centric Event Representations at the Document Level and Beyond


Shaotong Sun: Measuring Data Access Latency in Large CPU Caches

Undergraduate and Graduate Highlights

Faculty and Staff Highlights

Alumni Updates

Photo Updates:

Alumni Updates (continued)



Mohammad Rafayet Ali, PhD ’20: My current employer is Colgate Palmolive Company.


Rahul Bhotika, PhD ’03: I am now Chief AI Officer for UnitedHealth Group.


Ritwik Bose, PhD ’20: I have left Knox College and joined as a Senior Researcher in NLP and Agentic AI at the Johns Hopkins Applied Physics Laboratory (Research and Exploratory Development Department).


Michael Chavrimootoo, PhD ’24: I started working as an Assistant Professor of Computer Science at Denison University (Granville, OH)!


Michal Jan Cierniak, PhD ’97: My current employer is Anthropic.


Myroslava Dzikovska, PhD ’04: I am now working at Ada Health, http://www.ada.com.


Frank Ferraro ’11: I’m now an Associate Professor in the Computer Science and Electrical Engineering department at the University of Maryland, Baltimore County.


Arsal Imtiaz, MS ’23: My current employer is Everlaw, Inc.


Benjamin Kane, PhD ’24: Since graduating, I’ve started a new job as an Applied Research Scientist at Openstream.ai.


Henry Kautz, PhD ’87: After retiring from URCS in 2022, my wife Christine and I moved to Charlottesville, Virginia, to be near our daughter and her husband. Colleagues at the University of Virginia convinced me to join the faculty there, which I am doing half-time. My current main project is developing an open-source platform for collecting and analyzing online behavior data from multiple sources to support healthcare research and clinical applications.


Guo Li, MS ’20: I am currently a Bedrock GenAI Software Development Engineer 2 at Amazon Web Services.


Pengchang Li, PhD ’17: My current employer is Nvidia.


Ronald P. Loui, PhD ’88: Spent five years researching the ages in the Torah, from an academic, non-religious perspective with newsworthy results. I had previously been interested only in Minoa and Indo-European migrations a decade ago. I believe I have a tight timeline using a few hints from Josephus fixing lunar year reporting errors, putting the whole story within the Amorite Hadad situation, and ready to publish, with all dates archaeologically correct and all ages biomedically plausible. Sarah was perimenopausal at 45, as my MD mother of notoriously high IQ once told me when I was 12 and not that interested. Still a miracle, but my wife turns 46 next month. The key is piecewise linearity in ages, and two planting seasons per annum in Northern Mesopotamia. Flood was a 1959 Karaca Dag tephra lahar on the Khabur, Cain was post-Sargonic, Abram born in Sanliurfa, Moses was Hyksos, lotta teenage fathers. Elsewhere, the startup is doing disinformation detection for the DoD and feeding companies like GROUND NEWS. I completed a four-year study on AI liability and have a new paper in a law book on the EU AI Directive, excoriating "indecent induction" and pointing to the chain of confidence, not just a chain of causality (UofR’s ex-President Seligman helped along this path). The key phrases here are "respondeat inferior" and "certification for extrapolation in a specific dimension." Happily booting linux from live usb a lot, still in awk, OS demos in c. In class I talk about planes without TMR falling out of the sky and the usability crisis. Kyburg’s sensor fusion problem for the USAF remains unsolved but I have better ideas now. I did catch F/A-18s in the air over Cleveland, finally, after consulting on the plane 35 years ago. At the time I was dating the daughter of the AsstSecDef who produced them, but didn’t know it: thought she was fun to play tennis with. Baseline rally, white skirt. Back in WashU days. Just caught a UAA women’s basketball match with Rochester visiting CWRU last Spring; been watching UAA WBB since 1990. Dog is so old now, gets a lot of steak.


Grigorios Magklis, PhD ’05: My new employer is AMD.


Lee Moore, MS ’81: In 2023, I retired from Xerox with 43 years of work in the research group and in product groups. My wife and I continue to live near Rochester (NY).


Brian Meeker ’07: I’m currently a senior software engineer at GetThru. Over the past year I’ve also gotten back into public speaking on the conference circuit, mostly on topics related to Elixir.


Sara Melnick ’11: I’ll be starting a new job this fall. I’m going to be a Lab Instructor in the CS department at Wellesley College.


Fabrizio Morbini, PhD ’10: I am self-employed at Unfinished Makerspace LLC.


Pranay Mundra, MS ’23: I have started my PhD at Yale Computer Science in the field of provably practical and accurate locally edge differentially private (LEDP) graph algorithms, under Professor Quanquan C. Liu.


Deepak Pandita, MS ’19: I joined the PhD program in CS at RIT last year.


Massimo Poesio, PhD ’94: As of the 1st of May 2023 I have a full professor position at the University of Utrecht in the Netherlands. And I got a project on disagreements in interpretation associated with that, discussed at:

https://ained.nl/en/computers-die-beter-snappen-wat-mensen-bedoelen-daar-houd-ik-me-al-zon-veertig-jaar-mee-bezig/.

Also, I have been working on dialogue recently, and in particular on conversational agents embedded in virtual world games such as Minecraft, and am organizing a Dagstuhl workshop on Human in the Loop Learning through Grounded Interaction in Games that will take place in December:

https://www.dagstuhl.de/en/seminars/seminar-calendar/seminar-details/24492.

On the personal side, I still live in Britain, and my son has now started A-levels, so two more years and he’ll be going to university - alas that won’t be Rochester, but perhaps for his Masters or PhD?


Ramprasad Polana, PhD ’94: My employer is now Credence Management Solutions LLC.


Matt Post, PhD ’11: In 2021 I joined the Microsoft Translator team as a Principal Researcher. I retain a courtesy adjunct appointment at Johns Hopkins University.


Daniel M. Russell, PhD ’85: Hi folks - I have changed employers... I’m now working for BOTH Stanford University and the University of Zürich (spending around 6 months in Palo Alto and then 6 months in Zürich). And completely enjoying it. 


Robert B. Schudy, PhD ’82: We now winter in Florida. We have a condo on Golden Isles Lake, where we have a dock for our sailboat. We love to cruise in the Florida Keys, and are planning to cruise to the Bahamas. I will shortly be Commodore of the Gulfstream Sailing Club, with whom we race and cruise. My wife Liz Watson is President of our condo association.


Uri Shani, PhD ’81: Married to Alexandra. My 3 sons are married and have 6 children altogether. Guy lives in Ann Arbor with 2 of my grandsons (Ofek is a CS student at UM, and Stav is at high school now). Erez who was born in Rochester lives in Tenafly New Jersey with 3 of my grandsons Neta, Eitan, and Ari, all are younger than Guy’s kids. My youngest boy Tomer lives in northern Israel with my youngest grandson Evyatar. All are well, thank God, me too.


Dongbin Suh, MS ’22: I am married and expecting a baby. I am also building a start up.


Chunqiang Tang, PhD ’04: I have been at Meta Platforms (a.k.a. Facebook) for 11 years. After spending a decade working on Meta’s private cloud, I transitioned to the area of AI and Systems Co-design early this year, working on Meta’s AI chip called MTIA, the Llama large language model, HPC for AI, ARM CPU, general hardware features like CXL, etc. This shift brings my work closer to my advisor, Prof. Sandhya Dwarkadas’s work on architecture, and I hope we’ll cross paths more often at ISCA. If you are curious, my home page is mostly up to date: https://sites.google.com/site/tangchq.

While my primary focus is Meta’s production systems rather than doing pure research, I continue to publish our advanced work as research papers. I am very proud of our accomplishments in research publications---our paper, “ServiceLab: Preventing Tiny Performance Regressions at Hyperscale through Pre-Production Testing,” won the OSDI’24 Best Paper Award, and another paper, “Contiguitas: The Pursuit of Physical Memory Contiguity in Datacenters,” was selected for IEEE Micro Top Picks ’24. Overall, I find great satisfaction in contributing to industry work that not only impacts billions of people but also advances the state of the art research in hardware, AI, and systems. 


Rahul Tripathi, PhD ’05: My employer is now Amazon, Inc.


Peter von Kaenel, MS ’94: I have been promoted to Vice President, Compression Technology at Harmonic. I am also a member of the technical Emmy award committee. 


Will Walden, PhD ’24: I defended in June and will be starting as a Research Scientist at the Human Language Technology Center of Excellence (HLTCOE) at Johns Hopkins University on October 1st. Among many other great colleagues, I’ll be working alongside Ben Van Durme (URCS PhD ’10). I also got married in May and my last name has changed to Walden.


Zichen Wang, MS ’23: My employer is now ByteDance Ltd. (TikTok).


Jeremy Warner, MS ’16: My employer is now Apple.


Erin Witmer, MS ’20: I am currently a Sr. Machine Learning Engineer at Pluralsight.


Jiupeng Zhang, MS ’20: My current employer is Amazon Web Services.


Share your outcomes and updates with the department!


ugalumni@cs.rochester.edu, gradalumni@cs.rochester.edu


And connect with us in the URCS Alumni Group


https://www.linkedin.com/groups/12655649/

News Bulletins

Ehsan Hoque’s Research on AI-Based Online Testing for Parkinson’s Disease is featured on NSF “Discovery Files” Podcast

Bit by Megabit

University of Rochester photo / J. Adam Fenster

Luke Auburn, Hajim’s Director of Communications, has composed an article featuring our department’s 50 years of history with a timeline of significant events in Computer Science.

Chenliang Xu speaks on Are video deepfakes powerful enough to influence political discourse?

In another article by Luke Auburn, Professor Chenliang Xu talks about deepfakes, targets, detection, and potential misuse.

Say “Meliora!”

By Ardash Kumar

If it isn’t clear from the graduation caps, the students in this photo are glowing because they just graduated from Rochester. They’re probably also pretty happy that they didn’t have to pay to have this photo taken. Thanks to Adarsh Kumar ’24 and the International Students’ Association (ISA), these and more than 40 other international students received high-quality photos of this major life achievement as part of ISA’s graduation photoshoot initiative.


Kumar took all the photos himself—in fact, the photo above is his second winning entry to an ISO photo contest. And yet, he is not a student of the arts. A native of Varanasi, India, Kumar chose Rochester to study artificial intelligence. “What makes Rochester great for AI is they offer a good amount and variety of AI electives, allowing me to complete my computer science degree with an AI concentration. In my country, I can only properly explore specialized tracks like AI while pursuing a master’s degree or PhD.”

URCS Students at the Pumpkin-Carving Contest. Photo by Neeley Pate