Towards a Single-Host Many-GPU System
- SBAC-PAD 2018
I work to create IT systems that push beyond the boundaries of what is typically possible today to allow us to solve problems that are beyond the reach of our current tools. Currently I am working to create AI Systems that allow clients to use their data to create models which bring value to their business processes, deployed in hybrid cloud. My work experience covers the design, creation and operation of very large, often distributed, computing systems. Early work includes the Andrew System (AFS, ATK) & DFS. I designed and managed the operational aspects of the largest SP computing system in IBM from 1995 to 2000. I designed & build the prototype system for GSA distributed storage system and designed and implemented of the low-level control and monitoring subsystem for Blue Gene/L and Blue Gene/P supercomputers. My team created the prototype bare-metal design which is now used in IBM's Cloud and is now creating secure hybrid cloud technology for use in public and private cloud data centers.