Trust the System: A Q M Sazzad Sayyed’s Mission to Make AI Safer
Portrait of A Q M Sazzad Sayyed. Photo sourced from LinkedIn.
A Q M Sazzad Sayyed, PhD’27, is an electrical engineering researcher at Northeastern’s MENTIS Lab, where he works on making AI systems safer and more reliable—ensuring they can be trusted when deployed in the real world. His work on robustness, out-of-distribution detection, and data integrity positions him at the forefront of a challenge that will only grow more critical as AI becomes further embedded in everyday life.
A Q M Sazzad Sayyed is pursuing his PhD in electrical engineering at Northeastern. Sayyed has been interested in physics since high school, but discovered a new interest in artificial intelligence while completing his bachelor’s degree in electrical engineering at Bangladesh University of Engineering and Technology. He always had a PhD in mind because of his devotion to learning, and it was through his undergraduate work with designing and improving neural networks that he realized he enjoyed research, too. While the Covid-19 pandemic delayed his PhD plans, it afforded him the opportunity to acquire some industry experience that would prepare him for his graduate pursuit.
While exploring graduate programs, Sayyed was adamant that he wanted to explore the practical application of artificial intelligence. A peer had recommended Northeastern, and when he dug deeper, he became interested in the “the state-of-the-art wireless research” happening there. From there, he discovered the timely research of the professor who would become his current supervisor, Assistant Professor Francesco Restuccia. Sayyed knew that Restuccia’s work integrating wireless communications and AI was something that he wanted to be a part of.
Research at Mentis Lab
Sayyed’s primary work at the MENTIS Lab centers on making AI systems more “robust and safe.” One thread of his research focuses on stability—specifically reducing the impact of unexpected inputs or interference that can cause systems to behave unpredictably. A key project within this area, ENCORE, improves out-of-distribution detection, enabling AI systems to detect and flag objects or scenarios they were not trained on. A second threat focuses on safety from within: developing methods to remove problematic data from an AI system’s training—a process that ensures the system does not retain information it should never have learned. His most recent work ties these threads together, improving overall reliability by helping AI systems recognize when they are likely to be wrong.
Sayyed presenting his research at a conference. Courtesy photo.
Beyond his research, Sayyed has valued the courses and seminars offered through the PhD Network, which have sharpened his research writing, communication, and leadership skills—tools he sees as essential for working at the intersection of technical and real-world application. That real-world dimension is central to why he finds this work so compelling. As AI systems become more embedded in everyday life, he believes the need for reliability and robustness will only grow: “something that is always going to be necessary—you need to have trust in the system.” For Sayyed, the research he is doing now is not just an academic pursuit but groundwork for the safe and practical deployment of AI in the world beyond the lab.
Advice and mentorship
As he nears the end of his PhD, Sayyed is candid about what the journey demands. Pursuing a doctorate, he says, “is a long game”—one that requires patience, constant reflection, and a willingness to sit with uncertainty. That last quality is perhaps the most important: when results don’t come as expected, Sayyed encourages students not to be discouraged, noting that negative findings often “lead to new insights that we were not even thinking about.” He also advises prospective PhD students to research labs carefully before committing to them, and to invest in building a strong relationship with their advisor from the start.
Several courses stand out to Sayyed for shaping not just his technical knowledge but how he thinks and communicates as a researcher. Professor Restuccia’s Deep Learning and Edge Computing in Wireless Networks, structured around student-led discussions, helped him learn to articulate complex topics confidently in front of peers. Distinguished Professor Josep Jornet’s Terahertz Communication pushed that further, challenging him to distill his work for a general audience—a skill he considers essential for any researcher. Assistant Professor Mahdi Imani’s Reinforming Learning course added another dimension, training him to approach difficult problems from multiple angles rather than defaulting to a single framework.
Appreciation and the future
Beyond his PhD, Sayyed hopes to build a career in academia—and Professor Restuccia’s mentorship is already helping him understand what that path requires. Through their work together, Sayyed has gained insight into the practical realities of academic life: mentoring students, securing research funding, and building the kind of outreach that sustains a research career long-term. Northeastern’s collaborative environment has been equally formative, connecting him with scholars from around the world at seminars and conferences and expanding a network he expects to draw on throughout his career. At the heart of it all is a drive that has defined his time at Northeastern—to keep learning, and to pass that knowledge on, advancing the field of AI for cyber-physical systems one insight at a time.