Apr 2015 – Apr 2017 · 2 yrs

• ROC Platform V4
• Designed and led the development team ( Senior Architects and Developers) for the next generation Big Data platform.The Data platform packages extremely scalable components at the Data Ingest(ETL),Data Processing , Data Access and Data Management layers.
• The platform caters to a mix of Real time streaming , Micro batch and Batch processing workloads serving multiple Use cases such as Reporting, Aggregation, Rule engine, Profiling and sets the foundation for the journey towards packaged Machine Learning and Analytics.
• The platform uses Apache Spark streaming , Spark SQL, Apache Kafka , Hive (ORC), HBase, Presto, Flume,Solr. Focus has been on the Real time aspects including complementing the Spark streaming ETL layer with a complete In memory distributed cache/data grid ( Apache Ignite) to achieve massive scalability at the ETL layer.
• Designed and led the development team for the Operational Analytics platform for Monitoring and collecting performance and functional metric based KPIs and centralized Log aggregation. The platform is based on Solr for the Data management layer and has a flexible Data visualization layer, monitoring and alerting capabilities with an extensible check based architecture.
• Transformed the Engineering Process to a complete DevOPs style approach ,The new ROC 4 platform is completely setup on a CI and CD pipeline using the following components
• ○ CI pipeline including fully automated UT,IT, SonarQube Metrics using Jenkins as the automation build framework.
• ○ Docker for packaging and release management.
• ○ Automating the CD pipeline to deploy the Docker images on an Openstack based Virtual clustered environment.
• ○ Ensured the Testing strategy incorporates both Functional and Non Functional Automation for various modules of the Platform and helped modularize the Automation suite into individual component level Suites to speed up execution and better coverage.

Sep 2013 – Mar 2015 · 1 yr 6 mos

• Platform TCO Reduction (ROC Version 3)
• Reduced the TCO for the application by avoiding dependency on the DB for lookups and enrichments by introducing a extremely scalable distributed cache layer at the App tier level. This helped reduce the TCO by 50% by avoiding expensive DB licenses.
• Tuned the Distributed cache performance to perform 3x better than the standard DB based lookups for the heavy data sets like the complete Subscriber and Account Data sets ( close to 240 M records in the cache)
• Introduced support for a true columnar MPP DB - HP Vertica at the Data management Tier to speed up the querying performance drastically and reduce TCO by shifting to a commodity server based Scale out model.

Oct 2011 – Oct 2012 · 1 yr

• • Review new feature development across product road maps to ensure it aligns to performance goals
• • Review Deployment and debug performance problems on production systems
• • Proficient in Use of various Profiling tools for understanding the Top CPU hogging code and suggest workarounds
• • Suggest deploymemt changes to better utilize the available resources
• • Evaluate Product fitment on Partner offerings
• • Managed multiple benchmark projects for Revenue assurance and ROC Fraud Management
• • Fine tune legacy code for achieving scalability
• • Involved in Hardware sizing of the product for very high loads.
• • High load benchmarking with Oracle Exadata and maximizing the performance numbers by utilizing features of Exadata (HCC, Flash Cache, Predicate push down) to suit the Product workloads.

Jul 2010 – Sep 2013 · 3 yrs 2 mos

• Achieved 40% reduction in Total Cost of Ownership for the product deployments by avoiding/reducing dependency on database and high speed storage subsystems.
• Project planning and execution of the entire TCO reduction architecture change from the conceptualization/design phase to project implementation to release phase.
• Hands on in designing and implementing a real time Telecom call detail record processing product based on a In memory architecture.
• Fully aware of the challenges with designing for a real time record processing system with huge volumes.
• Took initiative to make the product architecture more white box testable by introducing tools to expose the In memory state and synchronization tools to control the synch flows from testing.
• Closely worked with Testing to incorporate new processes and automation suites to ensure a wide coverage of the new In Memory Architecture
• Closely worked with and supported the delivery manager for the beta customer/first implementation of the product till production launch.
• Closely work with Bid Management to answer big sized RFPs

Mar 2006 – Jul 2010 · 4 yrs 4 mos

• Integrated the ROC Revenue Management and ROC Fraud Management products using a “Common Data Model”.
• Changed the product functional architecture to a generic architecture supporting flexible configurable multiple aggregation types and data streams for Fraud monitoring.
• Changed the presentation layer from a thick client to Web based client based on Ruby on Rails and EXT JS.

Mar 2002 – Mar 2006 · 4 yrs

• Worked closely with the CTO to change the product architecture from a monolithic multithreaded application to a distributed architecture involving multiple processes & services.
• Conducted a organisation wide training for developers and senior developers on “Advanced C++ - Templates , The Standard Template Library and Boost Libraries, Working effectively with Legacy code”