How One Computer Scientist Is Rethinking Sustainable Computing at Scale
Content
“I started my PhD defense with one statement: climate change is real,” says Dr. Walid A. Hanafy. As a researcher in computer science, Hanafy understood the impact computing has on climate, and focused his dissertation research on developing sustainable computing systems that aim to reduce the environmental impact of our digital world.
Now a postdoctoral research associate in the Manning College of Information & Computer Sciences at the University of Massachusetts Amherst, Hanafy is helping lead a new wave of research at the intersection of computing and climate.
Hanafy works alongside his advisor, Prashant Shenoy, PhD, a leading researcher in sustainable computing and distributed systems. Under Shenoy’s guidance, Hanafy began exploring how to quantify and reduce the carbon footprint of digital computing systems. This is an emerging area of research that sits at the intersection of cloud computing, energy systems, and climate science.
From Egypt to UMass
Originally from Egypt, Hanafy earned his bachelor's and master's degrees in computer engineering from Helwan University. When considering PhD programs, he looked for two key factors: strong research in machine learning systems and the opportunity to work on integrating efficiency in them. Hanafy says, “When I saw Professor Shenoy’s work in sustainable computing and energy-efficient machine learning, I knew I had found the right place.”
At UMass, Hanafy became part of a deeply collaborative lab culture. “Everyone works with everyone,” he says. “You’re encouraged to share, collaborate, and learn. That’s rare—and it makes a huge difference.” His collaborations extend beyond the Shenoy lab to include UMass professor of Electrical and Computer Engineering David Irwin, and professors of Information & Computer Sciences Ramesh Sitaraman and Mohammad Hajiesmaili. He credits this collaborative interdepartmental culture with helping him see computing as a powerful tool for addressing complex, real-world problems.
The Hidden Cost of Energy
To understand Hanafy’s research, start with a simple idea: every watt of electricity we use comes at a carbon cost. “Not all energy is created equally,” he explains. “Electricity from coal is high in carbon emissions, while renewables like solar or wind are much cleaner. But most electricity grids use a mix—and that mix changes depending on the time and place.”
This insight is at the heart of carbon-aware computing, one of Hanafy’s key research areas. By understanding the carbon intensity of the energy available at a given moment, computing systems can shift tasks to draw energy to when and where the energy is greener. For example, researchers can process data in a region with cleaner energy production (spatial shifting) or delay computing (temporal shifting) until a different time of the day when green energy is more abundant. Temporal and spatial shifting allow computing systems to run more sustainably, without compromising performance or reliability. Hanafy says, “We're not telling people to stop watching videos or sending emails. We're just asking: Can we run these tasks at better times, or in greener places?”
“There’s no free lunch in sustainable computing, but once you know the trade-offs, you can design systems that make better choices.”
Shifting the Paradigm: From Energy to Carbon
For years, computing systems were optimized for energy efficiency. That made sense as less energy use meant lower costs. But, Hanafy argues, we’ve reached a ceiling. He explains, “Data centers now operate near perfect Power Usage Effectiveness. Efficiency gains are flattening, and demand is still growing. Energy efficiency alone isn’t enough.” That is why his research is part of a growing movement to treat carbon as a first-class design metric.
Unlike energy, carbon directly reflects environmental impact. The key challenge is that carbon-aware computing often changes the business as usual, thus comes with carbon vs. cost, carbon vs. performance, or carbon vs. energy trade-offs.
Hanafy says, “There’s no free lunch in sustainable computing, but once you know the trade-offs, you can design systems that make better choices.”
Beyond Carbon: Edge AI and Resilient Systems
While carbon-aware computing is a core part of his work, Hanafy also tackles other angles of computing efficiency. One of his projects focuses on building resilient and energy-efficient machine learning systems for resource-constrained environments, such as edge devices operating in rural areas or under unreliable network conditions.
Here, the challenges are different. Systems must be designed to withstand failures, adapt to dynamic workloads, and operate with limited energy or memory. Hanafy has developed strategies for failure resilience, smart model placement, and adaptive performance tuning.
Whether it's carbon-aware scheduling or failure-proof AI, his work converges on a central theme: how to design computing systems that do more with less in a sustainable way.
Projects with Real-World Impact
Hanafy’s award-winning paper, CarbonScaler: Leveraging Cloud Workload Elasticity for Optimizing Carbon-Efficiency, which earned Best Student Paper at ACM SIGMETRICS 2024, was a turning point in his career. Hanafy notes, “That paper opened up a lot of new directions while helping to identify key challenges to push towards more ambitious goals.”
Building on that foundation, Hanafy has contributed to major sustainable computing research initiatives with real-world reach, including CarbonFirst and the National Science Foundation’s Computational Decarbonization Expedition (NSF CoDec). CarbonFirst aims to improve the carbon efficiency of cloud infrastructure that powers technology from social media to artificial intelligence. That project laid the groundwork for CoDec, a larger, $12 million NSF Expedition led by UMass in collaboration with five other universities. CoDec takes the research a step further, applying carbon-aware principles to societal infrastructure, including buildings, cities, and transportation.
Looking Ahead
The demand for computing is only accelerating, and so is the urgency to ensure it serves a more sustainable future. At UMass Amherst, researchers like Hanafy are not just shaping what the future of computing looks like, but also addressing the pressing need for more responsible, resilient, and environmentally conscious systems.
Through innovative projects and cross-campus collaborations, Hanafy’s work highlights how research can move beyond the lab and into the world, impacting everything from cloud infrastructure to climate policy. It also reflects what makes UMass Amherst distinct: a place where technical excellence meets real-world purpose, and where students are empowered to tackle the defining challenges of their generation.
Written by Marissa Hanley, PhD student in Plant and Soil Sciences, as part of the Graduate School's Public Writing Fellows Program.
This article originally appeared on the UMass Graduate School's website.