651 63119


Assistant Professor

  • Ph.D. (Computer Science, École Polytechnique Fédérale de Lausanne, Switzerland, 2015)
  • M.Sc. (Computer Engineering, University of Belgrade, Serbia, 2010)
  • B.Sc. (Computer Engineering, University of Belgrade, Serbia, 2007)

Djordje Jevdjic received his bachelor (2007) and master degrees (2010) in computer engineering from University of Belgrade, Serbia as the top student in the class, receiving numerous awards from the University of Belgrade, the Serbian Association of Professors and Scientists, and the Serbian government. During his master studies, he worked as a full-time teaching assistant for a variety of courses in the Computer Science and Engineering department. Djordje received his PhD degree in 2015 from the School of Computer and Communication Sciences at EPFL, Switzerland, where he received an EDIC doctoral fellowship from the department and worked in the Parallel Systems Architecture Laboratory with Prof. Babak Falsafi. Djordje's primary research interests are in computer architecture, with an emphasis on increasing the efficiency of server systems for data centers to support sustainable growth of IT services. He believes in holistic approaches to application-system co-design, which include tighter integration of applications, systems, and emerging technologies, and rethinking of the established abstraction layers. Djordje's PhD research focused on understanding of conventional database and modern big data applications and the implications they have on the architecture of server processors, in particular on the memory systems. Djordje is one of the primary architects of CloudSuite, the first publicly available open-source benchmark suite whose purpose is to help with the design and evaluation of moderns servers. CloudSuite has been extensively used by renowned academic and industrial (both research and product) groups to drive their research and design and has been integrated into Google's PerfKit Benchmarker. Using CloudSuite, Djordje demonstrated severe inefficiencies of modern servers when executing big data applications, for which he received the best paper award at ASPLOS in 2012. To address the identified inefficiencies, Djordje looked at the emerging memory technologies, such as die-stacked DRAM, as a way to mitigate the memory wall faced by server processors. For his work on die-stacked DRAM caches, which was published in top computer architecture conferences, he was awarded the Intel Doctoral Student Honor Fellowship in 2013. Upon completion of his PhD, Djordje received a fellowship grant from the Swiss National Science Foundation for postdoctoral research at the University of Washington, where he worked in the Paul Allen School of Computer Science and Engineering with prof. Luis Ceze and Dr Karin Strauss. His postdoctoral research was focused on efficient and secure storage of multimedia using emerging storage technologies, as well as on reviving the processing-in-memory paradigm. In 2017 Djordje joined the DNA storage team at Microsoft Research in Redmond as a visiting researcher, working on approximate storage of videos in DNA. During his stay at Microsoft, Djordje created the first fully automated pipeline for encoding and decoding data to and from DNA, and created an efficient and distributed error analysis pipeline running on Microsoft Azure, all of which are extensively used in production at Microsoft. Djordje's most notable publications include ASPLOS 2012 (best paper award), ISCA 2012, ISCA 2013, MICRO 2014, TopPicks 2014, ASPLOS 2017, NIPS 2017 (selected for NIPS highlights).


  • Memory and Storage Systems for Big Data

  • Approximate Computing

  • DNA-Based Data Storage

  • Systems for AI

  • 3D-Integrated Circuits

  • Systems Modeling and Evaluation






  • 2012 - Received the best paper award at ASPLOS

  • 2013 - Awarded the Intel Doctoral Student Honor Fellowship

  • 2015 - Postdoctoral Fellowship by Swiss National Science Foundation


Introduction to Operating Systems