Amir Ali Ahmadi, Professor of Operations Research and Financial Engineering
Ahmadi has been a professor at Princeton University’s Operations Research and Financial Engineering Department since 2014 and is an affiliated faculty member of the computer science and mechanical and aerospace engineering departments and the applied and computational mathematics program. He is also a participating faculty member of the Center for Statistics and Machine Learning (CSML).
Before coming to Princeton, Ahmadi was a Herman Goldstine Fellow at the IBM Watson Research Center from 2012 to 2014. He was a postdoc at the Massachusetts Institute of Technology’s (MIT) Computer Science and Artificial Intelligence Laboratory and the Robot Locomotion Group, Laboratory for Information and Decision Systems from 2011 to 2012. He earned two bachelor degrees at the University of Maryland, one in electrical engineering and the other in mathematics. For his master’s degree, he went to MIT where he studied electrical engineering and computer science. He also earned his doctoral degree in electrical engineering and computer science (with a minor in mathematics) from MIT in 2011. His thesis was titled, “Algebraic Relaxations and Hardness Results in Polynomial Optimization and Lyapunov Analysis.”
During his academic career, Ahmadi has had numerous papers published and won awards in teaching and research.
Ahmadi’s main research focuses broadly on the science of optimization, a discipline essentially of “making the most of every situation,” he said.
“We all do optimization everyday: how you choose to commute to your job from home, for example. You make your choice to minimize travel time, or maybe the cost of your commute, or a combination of both,” he said.
A situation that needs to be optimized (also known as an optimization problem) has three elements, he said. These elements are the following: A decision variable which means a choice you make among several alternatives, such as multiple routes to work or different means of transportation; an objective or goal where you want to minimize or maximize a quantity such as commute time or cost; and the third element is constraints, which is any fact of life that limits your choices, for example traffic lights or train schedules.
Beyond mapping a route from home to work, optimization appears in larger-scale decisions such as what items to keep on stock as a massive retail store, or how to invest a million dollars among 200 stocks as a hedge fund, Ahmadi said.
In Ahmadi’s research, he and his students look at the mathematical underpinnings and theories behind optimization and design algorithms that aim to find the best solution to such problems. When it comes to CSML, Ahmadi views machine learning and the other tools being taught at the center as an essential component of modern optimization techniques.
“In machine learning, you have data and you want to use it to design computational gadgets that make predictions about the future. The task of finding the best gadget from data is an optimization problem,” he said. This simple fact has brought the optimization and machine learning communities very close together.
“What’s exciting about CSML is that it is bringing different departments together and it encourages interdisciplinary research,” he said. “I have also been involved in overseeing the undergraduate and graduate curricula of the center under Peter Ramadge, the CSML director.”
“It’s stimulating for me to be involved at CSML because I get to see all the teaching and research activities that my colleagues are pursuing in various fields related to machine learning”, he continued. “It feels like the campus is coming together, not just within engineering, but also the social sciences.”