May 2015 – Apr 2017 · 1 yr 11 mos · Bengaluru Area, India

• ★ Tiered Data Store
• The aim is to create a datastore which stores the transactional data in Mysql and move the older data to a secondary store like HBase.
• Created a datastore using Sharded Mysql, HBase and Real Time Change Propagation System. This system is useful for the scenarios in which the Entity has a short lifecycle after which there are no transactional use-cases for it but it should be queryable for historic use-cases.
• As a part of building this system we leveraged the existing open-source data sharding framework
• Vitess which is used by youtube to scale their traffic and also contributed back to the Open Source.
• Sharding is transparent to the application layer
• Entity Graph Lifecycle Management ensures that the data is purged from the transactional store once its lifecycle gets completed.
• For more details on the feature set and general understanding check out https://github.com/flipkart-incubator/polyglot
• Explore Vitess @
• http://vitess.io/overview/
• https://groups.google.com/forum/#!forum/vitess
• https://github.com/youtube/vitess

Mar 2012 – Aug 2014 · 2 yrs 5 mos · Hyderabad Area, India

• Worked on Multiple Projects:
• ★ Query Explorer Platform:
• Worked on an In-memory Bipartite graph, which is used to serve the Query/URL Click information for last 1 year queries served by Bing.
• Extended the Bipartite Graph to ingest Query/Entity information as well as to find out similar queries[Queries which have similar intent]
• Worked on building the infrastructure and implementing some functionalities, which makes the following scenarios to trigger within few seconds:
• Find out related entities for a query
• Find out all queries related to an entity
• Find out similar queries(queries with similar intent) for a given query
• This Graph is used to serve the information within few seconds, which were used by the Data Scientists to perform various experiments
• Worked on rewriting the way the original Query/Url Graph was being built to make it stable and scaling it horizontally.Also worked on optimizing the space that this graph takes on M/R file system.
• ★ Athena:
• Worked on building a web tool to query information from a key-value store
• This tool was used extensively by different teams to verify the correctness of data ingested in key-value store.
• This tool was the single source of blacklisting/whitelisting queries which are being ingested in the key-value store.
• ★ Data Insights:
• Worked on a tool which will help in query classification
• The aim was given a list of queries, it classifies different queries according to their domain intent.
• The features includes, showing query classification, session data and labelling different queries.

Jun 2010 – Mar 2012 · 1 yr 9 mos · Bangalore

• ★ Pagelet Producer:
• I worked on this Product since November 2010.The product is basically a proxy and security Management system used for exposing internal applications to external users.
• Owner of the Following features: OpenSocial Integration, Hosted Content Editor and WSRP/JPDK Producers.
• During my work at ORACLE
• I Contributed to the Design and Development of the automation framework for Pagelet producer.
• Worked on the Development of Gadgets using Gadgets and OpenSocial API to check the integration of Open Social feature with Pagelet Producer.
• Worked on REST Framework and API testing using Junit.

May 2009 – Jun 2009 · 1 mo · Gurgaon, India

• I worked as a Summer Intern in Vizexperts on implementing the CUDA(Compute Unified Device Architecture) backend for Brook++ language.
• I implemented and added the support of some features like Domain , Domain of Execution, Reduction and Graphics Interoperability(D3D9, D3D10 and OPENGL)for Brook++.The feature were designed to add the support of Brook++ to Nvidia GPUS.