Designed and prototyped distributed SQL queries for SQL Server. The functionalities included pushdown predicates, partial aggregation, and scale on demand.

Worked extensively with different teams in Microsoft to analyze and combine different products in SQL to achieve distributed SQL queries for customers.

▪ Designed and created end to end process with both client and server that communicates with Google protocol buffers. Main functionality include supporting concurrent read and write requests and file logging to verify correctness. These were all tested extensively with millions of requests with thousands of clients.

▪ Improved the stub cache used for all processes, which is a thread-safe cache that stores network connections for local processes that need to communicate to remote processes. The stub cache load balances the requests and the bandwidths

of the TCP endpoints, so that the local load distribution is optimal.

▪ Isolation testing of the stub cache improvement with up to 100% increase in throughput and negligible effect on latency.

▪ Researched on new compression algorithm LZ4X with extensive benchmark testing.