I'm responsible for the storage engine and other "close to the metal" components of the Spanner distributed database system. Spanner is the back-end for most of Google's billion-user products, and serves over 2 billion QPS on over 6 exabytes of data. The storage engine consumes the bulk of Spanner's total resource costs, and by itself is one of the largest systems at Google -- so optimization is a large part of what we do. We also drive development of novel database features, taking advantage of our expertise in unique opportunities in the storage layer.
My team is also responsible for Spanner's response to hardware Silent Data Corruption (a.k.a. mercurial cores) and for database optimization for new hardware (including ARM chips, accelerators, and new memory technologies).