Working on Spanner, Google's globally-distributed synchronously-replicated database.

Proposed, formed, and led a team of 3+ engineers, working on data pipeline matching signals from 10+ sources. Designed a migration of >300TB of data from Bigtable to Spanner.

Built client component for the world’s first in-memory execution engine for Hadoop MapReduce, which improved performance by up to 50x by eliminating disk IO and intermediate key sorting.

Implemented data grid components: database persistence connector, custom serialization framework, authentication, bulk operations, and advanced object querying.