Mohammad Javad Amiri, an assistant professor in Stony Brook University’s Department of Computer Science, has received a National Science Foundation CAREER Award to develop distributed data management systems that automatically adapt to changing workloads, failures and potential attacks.
The five-year, $631,659 award will support his project, “An Adaptive Ecosystem for Distributed Transaction Processing in Untrustworthy Environments,” while expanding research and educational opportunities for Stony Brook students.
Amiri’s research addresses a challenge facing services ranging from banking to healthcare: how to keep data processing fast and reliable when it involves many computers, often across organizations that do not fully trust one another. His project will use reinforcement learning to help these systems monitor changing conditions and adjust how they coordinate and process transactions.
“The problem I’m solving is how to make these systems adaptive, being able to detect when something is wrong or changed and automatically reconfigure themselves to stay fast and reliable,” Amiri said.
The award is part of NSF’s Faculty Early Career Development program, which supports early-career faculty who integrate research and education. For Amiri, it provides sustained support to bring years of research into one integrated system while preparing students to contribute to the field.
“Amiri’s research addresses a critical challenge: keeping distributed systems reliable as the conditions around them change,” said Samir Das, professor and chair of the Department of Computer Science. “This CAREER Award will help him advance that work while giving our students opportunities to contribute to research with broad applications across industries.”
Distributed systems underpin a wide range of services, including financial transactions, healthcare records, supply chains and cloud computing. But the computers participating in those systems can become overloaded, lose connectivity or behave maliciously. According to Amiri, systems built around a fixed set of operating conditions can struggle when those conditions change.
His project will develop an adaptive system that monitors workload patterns, network behavior and potential attacks, then uses reinforcement learning to select operating strategies suited to those conditions. That could mean changing the protocol computers use to reach agreement on transactions, adjusting computing resources or reorganizing how data is partitioned across the network.
The research spans three connected areas: adaptive consensus, which addresses how computers reach agreement; adjustments to hardware resources and transaction management; and changes to how data is partitioned and how groups of computers communicate. Together, these efforts aim to help the system respond to changes across multiple parts of its operation.
The project draws on consensus protocols, scalable data-partitioning systems and learning-based techniques that Amiri and his collaborators have developed over several years. It also extends the ideas explored in his book, Blockchain-Enabled Large-Scale Transaction Management, which examines how to process transactions at scale while balancing performance, scalability, confidentiality and verifiability.
Where the book explores the strengths and trade-offs of different architectures, the CAREER project asks how a system can learn which architecture to use as conditions change.
“The book is the map of the design space, and the CAREER project is about building a system that navigates that space automatically,” Amiri said.
The award will allow his lab to pursue that integration over five years, with sustained support for doctoral students to take ownership of the project’s research areas. Undergraduates will also have structured opportunities to participate, gaining experience with distributed systems research earlier in their education. The project builds on existing collaborations with researchers at the University of Pennsylvania, UC Santa Barbara and UC Davis.
Amiri plans to bring the research directly into the classroom by expanding course offerings in distributed systems and data management at Stony Brook. Students will learn the adaptive, learning-based techniques being developed in the lab and have opportunities to apply them to real research problems.
Beyond the classroom, the project includes outreach aimed at broadening participation in computing and helping people develop new workforce skills in distributed systems and artificial intelligence.
For Amiri, receiving the award affirms a research direction he has pursued since his doctoral studies. It also recognizes the contributions of students and colleagues who have helped advance that work.
“Personally, it’s gratifying because so much of this has been a collaborative effort with students and colleagues over many years, and a CAREER award feels like recognition of that shared work as much as anything I’ve done individually,” Amiri said. “CAREER awards are as much about the next generation of researchers as the research itself, and I take that mentorship charge seriously.”
Amiri’s previous accomplishments include co-authoring Blockchain-Enabled Large-Scale Transaction Management, published by Springer in 2025, and the paper “The Bedrock of Byzantine Fault Tolerance,” which received an Outstanding Paper Award at USENIX NSDI 2024. His research on distributed systems, data management and blockchain technologies has also appeared at leading conferences including SIGMOD and VLDB. Before joining Stony Brook in 2023, he was a postdoctoral researcher at the University of Pennsylvania, following his doctorate in computer science at UC Santa Barbara.
Add comment