I am a research engineer in the Cloud Efficiency group. Currently, I am working on the Power Capping and Harvesting project.
I obtained my PhD from Duke University; my dissertation was on how to leverage emerging datacenter capabilities, such as precise synchronized clocks and software-defined storage, for optimizing protocols and improving performance of transactional key-value storage systems in a datacenter. In addition, I worked on scaling and managing tail latencies in distributed file systems running on harvested resources (CPU, memory and storage) in datacenters over two summer internships at Microsoft Research.
Prior to attending graduate school, I spent four years working at a storage startup called VeloBit and subsequently at HGST (through VeloBit’s acquisition). Among many things, I worked on similarity detection and delta compression of data for reducing writes and increasing lifetime of NAND flash-based Solid State Drives (SSDs). I also lead a project for developing a distributed, block-level cache using virtualized SSDs at HGST.