À propos
Subru is a Principal Architect at Microsoft in the Gray Systems Lab (GSL) (opens in new tab) team, currently focusing on using ML techniques to build autonomous data systems in the cloud and improve cluster resource management efficiency at datacenter scale. Previously at Microsoft, Subru was a Principal Research Engineer working on different aspects of Hadoop YARN scheduling, specifically scaling it to 50K+ nodes and providing SLA guarantees. This work is a critical driver for the internal Cosmos BigData clusters having scheduled nearly one trillion tasks that processed more than a Zettabyte of production data.
Prior to Microsoft, Subru worked at Yahoo! where he contributed to Apache Oozie’s precursor, near real-time stream processing on Hadoop and HBase replication & compaction.
He is also a member of the Apache Hadoop PMC where he has been actively contributing since 2007 with emphasis on YARN resource management.
Subru’s research interests include autonomous data systems and large-scale distributed systems, specifically around resource management and operational efficiency.