Skip to Main Content
Elias Khalil joins MIE as an assistant professor. His research interests are in artificial intelligence, with a focus on machine learning and discrete optimization.

Elias Khalil obtained his PhD in Computational Science and Engineering from Georgia Tech (2019), an MS in Computer Science from Georgia Tech (2014), and a BS in Computer Science from the American University of Beirut (2012). His research interests are in artificial intelligence with a focus on machine learning and discrete optimization. Writer Lynsey Mellon spoke to Khalil about joining MIE, his research and his advice for U of T Engineering students.

What drew you to MIE at U of T and made you eager to accept a position here?

I chose MIE at U of T for a number of compelling reasons: strong research programs, excellent graduate students, a friendly community of faculty and staff, many potential faculty collaborators both in MIE and at the university level, ambitious plans to grow artificial intelligence within U of T Engineering, and a great, welcoming, diverse city.

What is the most memorable experience in your career so far?

Halfway through my master’s studies in the U.S., I started working on my thesis. The result I was seeking seemed out of reach at first; I was only starting out with research and the theory I was looking to derive was complicated. One of the most memorable experiences I’ve had so far in my career was proving that theorem. It took many iterations and really pushed my limits at the time, and I remain proud of that result. The thrill of making a dent on a research problem, which I first experienced then, remains the main driver for me.

Can you share a little about your research and what you like about it?

Much of my research centres around the following question: how do we design optimization algorithms that improve with experience? Much of our modern world’s operations boil down to solving optimization problems: assigning delivery trucks to routes such that transportation cost is minimized; managing inventory to minimize storage and service costs subject to varying customer demand; choosing which land parcels to conserve to maximize the probability of species conservation; allocating physicians to patients based on expertise and availability to maximize health outcomes.

Designing algorithms that work very well for a particular optimization problem can be a tedious task for us humans, both for theoretical and practical reasons: how should a truck-routing algorithm behave differently when used to optimize daily delivery routes in Toronto versus Montreal? The post office might need a PhD-level computer scientist or operations research specialist to analyze the data and modify existing algorithms to achieve improved performance in both cases. It is often easier to design algorithms that work OK on a class of optimization problems, but then that might not be enough for the end-user: we want really good algorithms so we can make better decisions, save on time and computing resources and solve more challenging problems.

This is where my research comes in: I design machine learning methods that leverage historical data to redesign or re-purpose optimization algorithms such that they perform better on a problem we care about. My research draws on a mix of techniques from computer science, machine learning/artificial intelligence and operations research, and I am interested in applications in supply chain management, urban planning and healthcare.

What do you hope to accomplish, as an educator and as a researcher, over the next few years?

In research, I am excited to work with students and collaborators to push the frontier in this new area of “data-driven algorithm design,” both on the theoretical and applied fronts. In education, I hope to show students the connections between data, algorithms, and applications and help them develop into multi-faceted engineers that can bring these technical skills into applications.

Do you have any advice for incoming students?

For undergraduates: concepts that may seem far removed from the real world can turn out to be really useful down the line if you put in the effort to understand them. For graduate students: read a lot of research papers and textbooks so you can quickly become an expert at a few select topics, and be nice to other researchers (let them know if you like their work!).

Do you have a favourite spot on campus or in Toronto?

I’ve only been in Toronto for a few weeks, but I already like the Back Campus Fields which often have a bunch of pick-up soccer games going every day during the summer.

Media Contact

Fahad Pinto
Communications & Media Relations Strategist