I am a Research Engineer (Member of Technical Staff) at
VMWare. I work at the intersection of software-defined
networking and storage systems. My work at VMWare is about designing
and implementing a data centric approach to software-defined data
center in our product NSX and improvising
it to work as a hybrid cloud.
My broad research interests lie in topics related to managing and processing very large amounts of data. I am also very interested in areas directly influenced by data storage and management such as data analytics and machine learning.
Earlier, I received my Ph.D. from Duke University's Department of Computer Science in 2015. I was part of the Duke Database Research Group. My Ph.D. work was on designing I/O efficient algorithms and data management techniques for Solid-State Drives. As part of this work, I developed efficient log-structured merge techniques for SSDs, developed efficient algorithms to maintain group-by's on random-access devices such as SSDs and cloud storage, and designed new concurrency control algorithms for indexes tuned to SSDs.
I have a wide range of experience in designing algorithms and storage techniques. My experience ranges from designing text ranking algorithms based on proximity of key word matches to classification algorithms such as decision trees and associative classifiers. Before I came to Duke, I developed a classifier called ACME which combines informative association rules in a statistically unbiased fashion. I have also developed two methods to construct decision trees, one about quickly constructing the tree (done using a prefix tree representation of the learning data) and a second one about constructing a high quality tree using association rules for increased accuracy in classification.