Sep 2022 – Aug 2024 · 1 yr 11 mos · Hyderabad, Telangana, India · Remote

Feb 2018 – Jul 2022 · 4 yrs 5 mos · Hyderabad Area, India

• I was the lead engineer on Gandiva (incorporated now as a module inside Apache Arrow), that provides a framework for LLVM based code generation for evaluation of columnar expressions. This was initially implemented for project and filter, and dramatically improved the performance of these in analytics use-cases. Worked on misc improvements in other operators and added support for decimal data types.
• Added Apache Iceberg support to dremio, to make it a first-class directory format supported inside dremio and provide interoperability between spark engines that create/alter the datasets and dremio clusters that provide analytic/ad-hoc query support on the same datasets.
• I was the prinicipal architect from dremio cloud, and re-architected the dremio execution engine into small, manageable micro-services that are multi-tenant and can evolve independantly. We used a novel model that has a mix of multi-tenant managed microservices and on-demand "compute instances" that can run in each customer's VPC.

Nov 2016 – Dec 2017 · 1 yr 1 mo · Hyderabad Area, India

• I worked on enhancements to "cold tier"for databases used by Amazon, to improve performance and reduce costs.

Jul 2009 – Oct 2016 · 7 yrs 3 mos · Hyderabad, Telangana, India

• I was one of the initial engineers in MapR - was part of a small team that developed a highly performant and distributed filesystem that could be used seamlessly with both hadoop and nfs. I worked on the directory operations, read-write codepaths, inode managements and replication/resync.
• I was the architect for MapR-DB, a low-latency distributed database and MapR streams, a pub-sub system built on top of the filesystem. Both of these were built for high scalability and fault tolerance, and had advanced capabilities like snapshots and cross-datacenter replication.

Oct 2008 – May 2009 · 7 mos

Sep 2004 – Oct 2008 · 4 yrs 1 mo

• Worked on several features on the Agami filer including Persistent Log, spilling the Persistent Log to disk, and supporting High Availability.

Jan 2002 – Jan 2004 · 2 yrs

• Integrated a third-party url filtering software SmartFiler 4.1 with NetCache.
• Worked on NetCache Object Store (for storing urls on disk) and data movement protocols.

Jul 2000 – Jan 2002 · 1 yr 6 mos

• Worked on porting Veritas Filesystem to Linux. Implemented the QuickLog performance feature and writeable snapshot feature for linux.
• Woked on porting Veritas Filesystem to AIX. Implemented the "Access Control List" feature and worked on issues related to filesystem operations and pagecache.