Dr. Amine Barrak is an Assistant Professor in the School of Engineering and Computer Science at Oakland University, Michigan. He earned his Ph.D. in Computer Science from the University of Québec, where his research focused on serverless architectures for scalable peer-to-peer machine learning. Dr. Barrak also holds an M.Sc. in Software Engineering from Polytechnique Montréal and a B.Sc. in Computer Engineering and Industrial Systems from the Higher Institute of Informatics in Tunisia. With a strong background in distributed systems, cloud computing, and machine learning, he has contributed to numerous high-impact research projects and published widely in top-tier journals and conferences. His professional journey includes multiple research internships and leadership roles in AI-driven initiatives. Dr. Barrak is also deeply committed to teaching, having delivered undergraduate and graduate courses on topics such as cryptography, cloud computing, and distributed systems. Recognized for his teaching excellence and scholarly work, he has received several prestigious awards, including the Faculty Research Fellowship and the Best Student Paper Award at CASCON. He actively mentors students, contributes to academic service, and serves as a reviewer for leading journals and conferences in computer science.
Dr. Amine Barrak’s research interests focus on distributed machine learning, serverless computing, and secure, scalable cloud architectures. He also explores IoT data analysis, NLP, and DevOps for resilient AI systems.
Resource-Aware Serverless Computing: Leveraging Stateless Paradigms for Cost-Optimized Cloud Applications
As cloud adoption continues to grow, so does the need for cost-effective and scalable application design. This talk explores how stateless paradigms, including serverless computing, stateless databases, and event-driven architectures, are being increasingly adopted to address performance, scalability, and cost challenges in modern cloud environments.
We discuss how resource-aware design helps match application components to the appropriate compute, memory, and bandwidth resources. The focus is on stateless application design, which avoids persistent state within compute components, enabling seamless scaling and distributed execution.
While stateless services offer many advantages, they also introduce challenges such as cold starts, repeated dependency uploads, and limited runtime control. To address these issues, we explore several optimization strategies, including code refactoring, reducing unnecessary complexity, and batch processing to minimize invocation overhead. We also examine how to orchestrate services efficiently in stateless environments using workflows and state coordination.
The talk concludes with a case study on distributed machine learning training using serverless computing, illustrating how stateless, resource-aware strategies can significantly reduce infrastructure costs while maintaining performance.