À propos
Adrian Caulfield is an active researcher in the computer architecture and systems communities. He received his Ph.D. from the University of California, San Diego in 2013 under the direction of Steven Swanson. He was a founding member of the Non-Volatile Systems Lab where he lead the Moneta project, using FPGAs to develop and prototype novel Solid State Disk (SSD) architectures targeting advanced memory technologies such as Phase Change Memory (PCM or PCRAM) and Spin-Torque Transfer MRAM. Achieving good overall system performance on these fast memory technologies required co-designing both the SSD architecture and operating system storage stack. He also worked extensively with Flash memory during his time at UCSD, designing well balanced, power efficient system architectures and characterizing the performance and intricacies of multi-level cell flash memories.
Adrian started at Microsoft in July 2013, joining the nascent Project Catapult as one of the lead researchers and developers. Project Catapult’s FPGA acceleration platform rapidly became central to Microsoft’s cloud strategy, enabling major advances in Bing ranking, cloud networking, and a number of other areas across the company. Adrian’s work on Catapult pioneered a number of low-latency, high-throughput networking technologies that have become core Microsoft’s large scale AI systems.
Adrian now leads the Specialized Networks and Prototyping (SNAP) team which drives AI network and system architecture, accelerator development and technology development and pathfinding.