Shahryar Zehtabi, PhD student in the Elmore Family School of Electrical and Computer Engineering

AI and machine learning are top of mind today, both in the pure technology space as well as in the countless real-world domains where they are being, or are envisioned to be, applied. Their enormous data sets and processing needs call for large numbers of compute resources — just witness the rash of new data centers being built — to segment the data and compute workload in order to most efficiently and quickly train, process, and run the models. This is often referred to as distributed machine learning, and Shahryar Zehtabi, a PhD student in the Elmore Family School of Electrical and Computer Engineering (ECE), is researching and developing innovative algorithmic solutions to advance this next phase of autonomous artificial intelligence.

What investigative avenues are you pursuing?

My research focus is on developing theoretically grounded algorithmic solutions for distributed machine learning. Thus, my work involves exploring new approaches to solve existing challenges prevalent in the machine learning community and analytically proving the correctness and efficiency of our proposed solutions. I believe that in moving from the Information Age to Industry 4.0, with its explosion of data, artificial intelligence will be dominating various aspects of our lives; therefore, developing fast and resource-efficient training mechanisms for AI is going to shape the technology of our future.

What spurred your interest in this subject?

Since becoming familiar with machine learning, creating autonomous artificial intelligence agents became my favorite technology due to its endless potential. As my undergraduate background was in electrical engineering with a focus on telecommunications, distributed machine learning was the perfect combination of both of my interests and my skills, i.e., machine learning and wireless networks. That is why I started my PhD, to pursue what is being called federated learning — where you train the learning models on a variety of data sources at individual sites to ensure proprietary integrity, then “federate” the outputs at a centralized server to build a single, aggregate model. This is one of the most recent promising advancements and avenues for research in distributed machine learning.

Why did you choose Purdue to continue your studies?

I chose Purdue for two main reasons among many others. First, Purdue's College of Engineering is widely recognized both inside the U.S. and internationally, and this was a deciding factor in my decision. More importantly, my research interests are perfectly aligned with my current advisor, Professor Christopher Brinton, and I thus became sure that Purdue was the best place

for me to pursue my PhD. The research mission of our lab, the Intelligence Optimization for Networks (ION) Lab, is to conduct research at the intersection of network science/optimization and machine learning, which is a perfect fit with my interests.

When did you first get interested in engineering and science?

I have been fascinated with technology since I was a kid. I remember being intrigued about how TVs and radios receive such precise signals, and also wondered how the internet worked. When I was seven, I used to sit in front of my father's computer for hours, trying to learn different software and tinkering with the hardware. As early as elementary school I decided to study mathematics and physics in order to uncover the science behind technologies that seemed mysterious to me. That is why I continued to pursue a dual major in electrical engineering and computer engineering in college.

What’s it like studying at Purdue?

I really admire working in the academic atmosphere at Purdue, getting inspired by the level of professionalism and hard work by the faculty, fellow graduate students, and the undergraduates at the school. Through countless hours that my advisor spends weekly with graduate students, we are able to learn how to become better researchers as PhD students. In the ION lab, we are also fortunate to work with some amazing and intelligent postdoctoral scholars, who give us invaluable advice in our research projects.

Beyond deepening subject matter expertise, what else have you learned at Purdue?

I believe that becoming an expert in a subject matter is only half of the overall knowledge acquired during graduate studies. Through interactions with my advisor, fellow graduate students, postdocs and internship managers, I was able to learn priceless lessons on better communication to make critical problem-solving decisions. Furthermore, I discovered that gaining better academic writing skills to present novel research ideas, both to peer scholars and to the general public, is as important as the research idea and the solution itself.

What’s the research environment like at Purdue? Opportunities to teach and publish?

I really appreciate the philosophy of Purdue's Elmore Family School of Electrical and Computer Engineering, which creates an environment for PhD students to focus largely on their research and not get tied down by other requirements. I was a teaching assistant for three semesters when I joined Purdue, two of them for an undergraduate class, ECE 301 (Signals and Systems). I enjoyed interacting with undergraduate students and seeing them navigate through the Signals and Systems material for the first time and learning it. I was also honored to receive the Magoon Award for teaching assistants in 2024. Teaching is something I truly enjoy, and I hope to have the chance to teach in the future alongside research. I have published two papers, one at the 2022 IEEE Conference on Decision and Control (CDC), another a spotlight paper at the 2025 International Conference on Learning Representations (ICLR).

What advice might you give to other students deciding where to attend graduate school?

My strongest advice for them would be to look into the research being conducted by the professors of their respective fields in different schools. If there is a professor whose research catches their attention, I would recommend they get in touch with them and see if their interests are aligned. With Purdue Engineering being among the top five in the nation, I am sure that graduate students can pursue their interests here and flourish in their fields with the help of the excellent faculty.

What about the future? What are your goals; what are you looking to accomplish?

My goal is to continue doing research in the field of AI and build my own team of researchers. I would like to experience this in both the research industry and academia, as I believe each of them have their own advantages. Tackling unsolved research problems is one of the most challenging tasks in life, and the challenge itself is a major motivation for me to pursue it. I hope to be in the forefront of AI research, and see its potential realized in advancing human life and experience in the future.

What are you up to in your spare time?

I like to stay active outside work, and enjoy playing sports, especially tennis and soccer. I recently started doing calisthenics as well, and I hope to learn even more physical training in sports skills.

Source: Purdue Engineering gradate student profiles: Shahryar Zehtabi