I studied engineering for my undergraduate degree at the University of Auckland in New Zealand. In my final year, I started thinking hard about what I wanted to do in the future, and as I had really enjoyed the research opportunities I had had to date, I decided to apply to PhD programs in operations research. After visiting the MIT ORC, I knew it was the best place for me.
My academic advisor was Professor Dimitris Bertsimas. My thesis focused on developing new algorithms and software for robust and adaptive optimization, and trying to develop an understanding of when and why these methods work best. I also worked on many side projects, including developing the JuMP modeling language, using machine learning to build better algorithms, participating in a variety of analytics projects, and creating two new classes. After graduation I'll be working as a research engineer at Google DeepMind, where I'll be solving cutting-edge problems in artificial intelligence.
I really enjoyed my time at the ORC. It was a lot of work, but the opportunities I had, the friendships I made, and the experiences I had made it all worthwhile. OR is not only at the heart of companies like Uber and Amazon, but it can also be found in hospitals, schools, and many other organizations. If you are interested in tackling problems with a combination of modeling, mathematics, and computing, then a PhD program in OR could be right for you—and there is no better place to study OR than at MIT.
Basic background information on yourself, perhaps where you came from, former education, etc.
I was born in Canyon, Texas but grew up in Saskatoon, Saskatchewan, Canada. Before coming to MIT, I attended McGill University in Montreal, and received my BA with Honors in Urban and Economic Geography in 1977. My goal at that time was to become a city planner.
What degree you received from MIT and when?
I came to MIT in 1977 as a master’s student in the Department of Urban Studies and Planning. That summer I met Dick Larson, who was then co-director of the ORC. This changed my life. At Dick’s suggestion, I pursued two masters’ degrees – in city planning and in operations research – and received both in 1979. I then spent one year working at Dick’s company Public Systems Evaluation as a researcher before returning to MIT as a doctoral student in urban studies in 1980 with a new goal of using operations research to study public policy problems. Given the importance of empirical argumentation in the policy field, I took several statistics courses in the mathematics department. In the fall of 1981, Hermann Chernoff pointed out that with only a few additional courses plus a thesis, I could earn a master’s degree in mathematics (which was really a degree in statistics). I took this advice and detoured from PhD study to earn my third MIT master’s degree in 1982. This led Dick Larson to call me a “master of all trades but a doctor of none.” However, I returned to his doctoral studies, and graduated with a PhD in urban studies in 1984.
Who your advisor was at MIT?
Dick Larson was my ORC and urban studies master’s advisor, and was also my PhD supervisor. Hermann Chernoff was my advisor for my masters in mathematics. Arnold Barnett was also influential, while in urban studies my committee included joint ORC faculty member Joe Ferreira, and Lang Keyes, an expert in housing policy.
What was the research you performed while at MIT?
My master’s thesis in OR and urban studies addressed model-based evaluations of criminal justice programs. I used OR models to re-analyze data produced from some then-current experiments and observational studies in policing and detention/incarceration. My master’s thesis in the mathematics department examined mixture models for the duration of stay in a large psychiatric hospital. Finally, my PhD thesis took an operations research view of public housing. New applicants are the customers, housing units are the servers, the duration of residence is the service time, while the tenant selection policies utilized by the housing authority create the queueing discipline. Unlike most of the systems we studied in classes like stochastic processes or urban operations research, reneging is an important fact-of-life in public housing queues. I developed several descriptive models for various tenant selection policies that estimated waiting times for new applicants and also changes in the demographics of housing projects over time, and also showed how these models could be used to design policies to achieve various social goals. I also modeled and solved an interesting scheduling problem now known as the relocation problem that originated with the modernization and redevelopment of public housing projects. Like a very large game of musical chairs, during the course of redevelopment, all tenants must be housed in apartments of appropriate size (and with appropriate features, e.g. for the disabled). Redeveloping buildings could only proceed if all tenants could be relocated successfully, but successful relocation relied on available vacancies elsewhere on site. This creates an interesting resource-dependent scheduling problem where the availability of resources depends upon the schedule pursued. My PhD thesis resulted in several journal articles in both OR (e.g. Operations Research, Management Science) and housing (e.g. Journal of Housing, Planning and Design) journals, while the relocation problem sparked several extensions and generalizations published by others in the computer science literature.
Where are you currently working?
I am the William N. and Marie A. Beach Professor of Operations Research at the Yale School of Management, Professor of Public Health at the Yale School of Public Health, and Professor of Engineering at the Yale School of Engineering and Applied Science.
What have you been doing since your OR degree?
After leaving MIT, I spent one year as a Research Associate at Harvard’s Kennedy School and two years in the Management Science Department at the University of Massachusetts-Boston before joining Yale in 1987. I’m now in my 30th year on the Yale faculty, but I have also enjoyed sabbaticals as visiting professors in many wonderful programs including Stanford’s Graduate School of Business, Columbia’s Graduate School of Business, the MIT Sloan School, the Technion-Israel Institute of Technology, the Hebrew University of Jerusalem, and UC Berkeley’s Survey Research Center.
Shortly after arriving at Yale, I began applying operations research ideas to policy problems in HIV/AIDS. Best known among these is the model-based evaluation of the New Haven needle exchange program, where by combining data obtained from the tracking and HIV-testing needles returned to the program with newly-designed models inspired by both malaria epidemiology and queueing theory, we were able to conservatively estimate that the rate with which new HIV infections occurred was reduced by 33% on account of the program. I worked on many other problems in the HIV/AIDS realm, including a project with the CDC that forms the basis for how HIV incidence is estimated in the United States.
After the terror attacks of 9/11 and the anthrax letters that followed, I was contacted by government officials who sought advice evaluating our nation’s readiness to withstand bioterror attacks. This led to wonderful collaborations with Larry Wein (former MIT OR professor now at Stanford) in the realm of emergency preparedness and response to smallpox and anthrax attacks. I advised Canadian and Israeli officials on these topics as well.
Repeated visits to Israel, initially centering on HIV/AIDS issues (I produced the first estimate of HIV infection rates in Israel back in 1994) but later the threat of bioterrorism, sensitized me to the operational issues involved in countering more conventional terrorism. This led to studies of the effectiveness (or lack thereof) of tactics to counter suicide bombings. I then began a research program focused on operational aspects of intelligence collection and analysis. For example, how many undercover agents are needed (and how should they be deployed) to catch as many terrorists as possible? Queueing theory plays an important role in this work, and that research is ongoing.
One more area of study -- initially as a diversion but later as an area of real interest -- I have applied OR ideas to sports. One example is how to fill out the draw sheet in a March Madness office pool; another is how to estimate the number of hockey wins individual players “create” in the National Hockey League.
Finally, having been in the OR world for more than half of my life, I am now functioning as an advocate for our field by serving as president of INFORMS for 2016, though I much prefer the title Member-in-Chief!
Advice/guidance to students coming to the ORC – why the ORC from your perspective?
For me, the ORC was about surrounding oneself with what seemed like the smartest people in the world, and together adopting MIT’s famous “can-do” attitude. Often when presenting a new idea to someone, you will get the reaction “that won’t work because…” At the ORC, the reaction was more often “that’s interesting – why don’t you try it, and while you’re at it, think about this…” I loved the courses, my professors, and my fellow OR students. I can honestly say that the ORC changed my life – and certainly for the better.
My advice to students is to stay open with your interests, and recognize that the techniques you are learning have application and utility in areas you probably have not even thought about. Of course you need to narrow your focus to complete your research program, but try develop a habit for looking at problems you know little about, and ask yourself how you would approach them as operations problems.
Another small piece of advice – take some time to learn about the history of our field. There is a reason why operations researchers tend to be expert in optimization and probabilistic modeling, as opposed to, say, number theory and topology. A mathematician looking at OR without any understanding of the evolution of the field might find the subset of tools we focus on to be arbitrary and even odd. But OR came from studying operational processes so as to improve them. That by itself explains our focus on optimization and stochastic models.
Finally, try to focus on the front end of problem identification and problem/model formulation. Often time pressure will push you into solving someone else’s (e.g. a funded advisor’s!) problem. But at least in my experience, figuring out the right problem to work on is both harder but also more valuable than the mathematical work that follows (and I would argue this is just as true for theoretical as applied work).
How did the ORC contribute to your development as a scientist and as a person?
What can I add? The ORC made me who I am. It gave me the tools to succeed along with the confidence that I could.
What is one memory about the ORC you carried with you?
The ORC faculty and students just came across as the smartest people in the world to me. No matter what you are doing, if you surround yourself with this caliber of people, many of whom will become your friends – well, you will have a hard time failing. I also remember a lot of blackboards filled with equations, diagrams, graphs – this was a place where people worked very hard, but it did not feel like work. It felt more like ground zero for the discovery of new knowledge. The ORC to me was the most exciting place I’ve ever been.
Any additional information you feel might be of interest to others…………..
I remain very proud of my MIT education. It has served me better than I could ever have imagined, and essentially everything I have achieved in life can in some way be traced back to it. Also, the ORC gave me the ability to continue learning, albeit in less formal ways.
Tell us about yourself.
I am originally from Belgrade, Serbia, and I completed my bachelor’s and master’s degrees at the University of Toronto, both in industrial engineering.
What degree did you receive from MIT and when?
I received my PhD in operations research (OR) in 2016.
Who was your advisor at MIT? What kind of research did you do?
My advisor was Professor Dimitris Bertsimas, and we worked together on two broad research areas. The first was on how to solve large-scale Markov decision processes using a new type of linear optimization formulation of the problem. The second was on how to make effective decisions in the presence of customers with product preferences—in particular, how to make strategic product line decisions under uncertainty about the underlying market, and how to make tactical product assortment decisions from limited data.
Where will you be going after graduation? What will you be doing? Why does it excite you to do this work?
I will be joining the Decisions, Operations and Technology Management (DOTM) group at the Anderson School of Management at the University of California Los Angeles as an assistant professor. I will be doing research, as well as teaching courses in operations management and analytics to MBA and PhD students.
I am really excited by this because I love research—thinking creatively about interesting and important problems is a lot of fun. There is something incredibly exhilarating when you discover a new theoretical result, or your simulations finish and some amazing insight is staring you in the face, or even when you sit down with a blank LaTeX document and turn it into a complete paper.
I also really enjoy teaching. At MIT, I have been a teaching assistant (TA) for The Analytics Edge class, in both its regular and executive MBA formats. The course teaches students how to use data to build powerful models for predicting things you wouldn’t think could be predicted—for example, how much a wine from Bordeaux will sell for at an auction, or how U.S. Supreme Court justices will vote on a case. I found it incredibly rewarding to be a TA for this course because it is a way to have an impact; each student who masters the material is someone who will take that knowledge and do something exciting and impactful with it tomorrow.
Why should prospective students come to the ORC?
The ORC is simply the best place to obtain a graduate education in OR. The faculty are leaders in the field and are extremely devoted to helping their students succeed and develop into the best researchers they can be. The students here are bright, highly motivated individuals, but as a group, they form a tight-knit, collegial, friendly community: There is an intramural soccer team; we go out for lunch together; we celebrate birthdays and housewarmings together, and so on. MIT, as an institution, is an energizing place to be—the people you meet around campus are talented and inspiring. Last but not least, the Boston/Cambridge area is wonderful—during my time here, I’ve really enjoyed exploring areas like Harvard Square and Brookline, and just like MIT, the attractions around MIT are first rate: For example, the best ice cream I have ever had in my life has been at Toscanini in Central Square.
What advice would you give to prospective students to the ORC?
Before you apply, do your research: Look up faculty who seem interesting to you, print out a couple of their papers, and read them in the summer before you apply. This will help you get a sense of what people are working on and whether or not you are interested in a particular area. Although it’s perfectly fine to be unsure about what you would be interested to work on, it always helps to have an initial idea.
How did the ORC contribute to your development as a scientist and as a person?
The only way to become a good researcher is to do lots of research. My graduate training at the ORC immersed me in research, and in all aspects of research. One of the most important skills that I have gained—and that I continue to develop—is choosing the right problem. Being able to solve a problem—analyze it theoretically, plan out and execute simulations, and so on—is one thing, but as an independent researcher, you have to be able to formulate the problem in the first place.
What is one memory about the ORC that you will carry with you?
There are many—one is the ORC retreat, which happens every year in late September. We rent out a house near a lake and spend the weekend together, enjoying good food, going canoeing/kayaking, playing football or soccer, and just having fun!
I was born and raised in Athens, Greece. I received my Diploma (a 5-year degree) from the Electrical and Computer Engineering School of the National University of Athens, Greece in June 1991. Just before I graduated I got a phone call from Prof. Bertsimas inviting me to join the ORC. And I did arrive in Boston in August 1991 and never left :-)
I graduated from MIT in 1995 with a Ph.D. from the ORC. My research dealt with queueing networks, developing distributional law approaches. I published a number of papers and I got the Second Place in the INFORMS George Nicholson Student Paper Competition in 1996 (a year before my husband - a LIDS student - got the exact same prize).
I am currently the CEO of Dynamic Ideas Financial, a financial technology company and a SEC registered internet advisor and a partner at Alpha Dynamics LLC, a registered independent advisor.
After graduating from MIT, I stayed one year as a post-doc and worked on Financial Engineering. At the time I was consulting for asset management companies and in 1999 I co-founded (with Dimitris) Dynamic Ideas and then sold the assets of the company to Amex. From 2002-2010 I was a member of the quantitative asset management group of Amex and then Ameriprise/RiverSource. I left in 2010, when I was a senior portfolio manager, and started Dynamic Ideas Financial and Alpha Dynamics.
The ORC (and before that the Engineering school) uniquely prepares students not only to understand the beautiful theory of optimization and stochastic systems but also to engage in real world applications. I have been using all the tools that I learned from ORC every day in my work, and often times in my family life. I love the ORC approach that every challenge is an opportunity for a well-designed and executed solution.
My memory of the ORC is that of a warm, multicultural, inclusive place. I have friends, older and younger, who studied at MIT and remember it as a hard place but the ORC was really a fun, friendly place to be. Our first Thanksgiving in the US we had no friends or family outside the people we have met at MIT and one of my classmates from the center, Alan Kaufman, invited all of us to spend the day with his family. It was such a kind act and it represented the spirit of the ORC community.
Peng Shi grew up in China and Canada, and studied at Duke University for his bachelor’s degree in mathematics and computer science. He obtained a PhD in operations research from MIT ORC in June 2016, under the guidance of Professor Itai Ashlagi, his academic advisor.
Shi’s PhD research was on prediction and optimization in school choice. In a school choice system, such as the one implemented in Boston, each student is given a list of public schools based on home location and ranks the available options in his/her order of preference. Schools also give priorities to various types of students. A centralized algorithm determines the assignment, while using the priorities and lottery numbers to break ties. Shi's research was on optimizing these assignment systems so that students have equitable chances to go to the schools they want, while the city's school busing cost is controlled. His research was applied to real data in Boston, and the Boston Public Schools implemented one of the plans he proposed in 2014. Various parts of his research were published in Interfaces, an operations research and management science journal.
After graduating, Shi will pursue a one-year postdoc at Microsoft Research New England, before taking a tenure-track faculty position at the Marshall School of Business at the University of Southern California. He plans to continue his research agenda in developing quantitative methodologies that benefit society, with a focus on optimization in matching markets. He will also be teaching classes in operations management. What excites him about this work is the opportunity to play with interesting mathematics while thinking about real-world problems, as well as the opportunity to interact with students.
Shi encourages prospective students to think about the years at graduate school not only as a time for intellectual and career development, but also as an opportunity to develop deep friendships and grow as a person holistically. In his five years at MIT, the friendships that he developed are what will remain with him the longest and will always be cherished.