Συντάχθηκε 22-07-2026 18:11
Τόπος: Λ - Κτίριο Επιστημών/ΗΜΜΥ, 145Π-58
Έναρξη: 23/07/2026 12:30
Λήξη: 23/07/2026 13:30
Who
Dr. Haralampos Gavriilidis
University of California, Berkeley
Abstract
Modern analytics increasingly spans heterogeneous data formats and systems rather than a single database engine. While this flexibility allows applications to combine the capabilities of different systems, it also creates fundamental challenges for query processing, interoperability, and efficient data movement. In this talk, we will examine how federated and composable data systems can address these challenges without replacing the engines that already work well. We will discuss XDB for decentralized cross-database query processing, XDBC for fast and adaptive data transfer across heterogeneous systems, and SheetReader for efficiently integrating spreadsheets into analytical pipelines. Together, these systems show how query execution, data exchange, and format integration can be redesigned to enable faster and more resource-efficient analytics across heterogeneous data environments. Finally, we will explore how these principles can be extended beyond relational data to support analytics across heterogeneous data, models, and systems.
About the Speaker
Haralampos “Harry” Gavriilidis is a Visiting Scholar at UC Berkeley EECS, working with Scott Shenker in the NetSys Lab, and a Postdoctoral Fellow at the International Computer Science Institute, supported by a DAAD IFI Fellowship. He received his PhD in Computer Science from Technische Universität Berlin, conducting research at BIFOLD and the DIMA group, advised by Volker Markl. His work focuses on making heterogeneous data systems operate efficiently as a unified analytical environment. During his PhD, he designed XDB, XDBC, and SheetReader, leading to publications at SIGMOD, VLDB, ICDE, EDBT, and Information Systems. He also led two research projects in collaboration with industry, including a German Software Campus project. Before entering academia, he worked for several years as a software engineer. His current research explores how declarative data processing can support analytics beyond relational data, across heterogeneous data, models, and systems.
Σύνδεσμος εκδήλωσης
