Sotirios Xydis

Sotirios Xydis is an Assistant Professor in the Division of Computer Science at the School of Electrical and Computer Engineering, National Technical University of Athens (NTUA), with prior faculty appointment at Harokopion University of Athens (2020-2023). He maintains ongoing collaboration with the Institute of Communication and Computer Systems (ICCS) since 2014 and previously served as an engineer at HEDNO (2015-2018) and postdoctoral researcher at Politecnico di Milano (2011-2013).

His academic credentials include:

• BSc in Electrical and Computer Engineering, NTUA (2005)
• MSc in Techno-Economic Systems, NTUA (2011)
• PhD in Electrical and Computer Engineering, NTUA (2011)

Dr. Xydis specializes in hardware/software co-design, energy-efficient hardware acceleration, and memory management for embedded and cloud-edge systems. His research bridges low-power circuit design, heterogeneous architecture optimization, and resource management frameworks, with particular emphasis on AI workloads and serverless infrastructures. Current projects target Edge AI accelerators (CONVOLVE), disaggregated memory systems, and LLM inference optimization through hardware-aware algorithms.

Analysis of his 2023-2025 publications reveals three dominant trends: (1) Energy-efficient hardware accelerators for Edge AI using approximate computing techniques, (2) Memory/resource management innovations for disaggregated serverless environments, and (3) GPU/FPGA optimization for LLM inference through dynamic frequency scaling and predictive throttling. These works consistently address the power-performance tradeoffs in heterogeneous computing systems.

His scientific recognition includes:

• Best Paper Award, IEEE/NASA/ESA AHS (2007)
• Best Paper Award, ACM PARMA (2013)
• Best Paper Award, ACM Computing Frontiers (2020)
• Hipeac Award at DAC (2019, 2020)

Dr. Xydis has secured over 15 European/national research grants as Principal Investigator and Technical Coordinator, focusing on hardware acceleration frameworks and energy-efficient computing. His advising encompasses graduate research in hardware design and optimization, though specific student names aren't publicly listed. Current projects include CONVOLVE for Edge AI and CollectiveHLS for collaborative hardware synthesis.

He is a core member of NTUA's Microelectronics Laboratory (Microlab) and collaborates with ICCS on hardware acceleration projects. His team develops frameworks like CollectiveHLS and throttLL'eM, with active participation in DATE, DAC, and ISCA conference communities.