Tushar S. Email & Phone Number

Redmond, Washington, US

• Built digital advertising supply and performance reporting data pipelines for advertisers and retailers (publishers). These pipelines were optimized to process terabytes (and petabytes in some cases) of ad-events data.
• Integrated reporting between Microsoft’s Bing Ads and Retail Media Marketing programs.
• Tuned data pipelines for optimum cost to performance ratio. Migrated data stack from AWS to Azure to private cloud.
• Identified and resolved bottlenecks, helped set processes to improve team’s productivity and code, services’ quality.

Redmond, Washington, US

New York, NY, US

• Designed and delivered data ingestion pipelines and tools for building MediaMath’s next generation Identity graph stack for cross device, household, and CTV ad targeting. This graph comprises of few billion devices.
• Building CTV Attribution to measure user conversions via impressions on CTV.

New York, NY, US

Building cross device ad targeting product, called Identity. Responsible for forming device -> user graphs, ingesting data from partners etc. Ensuring components are GDPR and Privacy Policy compatible.

New York, NY, US

Developed a Cross Device Impressions reporting tool based on Apache Spark Streaming. This tool gives campaign managers insights into how a campaign performed across multiple devices owned by same user. The tool uses cutting edge features of Spark, such as windowing, stateful updates, check pointing etc.

• E-Commerce Recommendation system: Built a custom collaborative filtering algorithm, and implemented it using Spark. This enabled the platform to provide shopping recommendations to consumers.
• Genome Sequencing using Hadoop: Implemented a genome sequencing algorithm using Hadoop MapReduce. This enabled us to horizontally scale otherwise very compute expensive sequencing processes.

Pune, Maharashtra, IN

• Apache Hive ODBC DriverPorted driver to Windows, developed and maintained ODBC APIs, secured database connection with TLS technologies, tuned driver to improve performance by 35% in case of import data scenarios.
• Sqoop Connectors for MS-SQL Server and MS-PDWPorted Apache Sqoop to Windows, evaluated different approaches and modified Sqoop core to support MS-SQL Server and MS-PDW.
• Hadoop distribution optimized for Intel® Xeon Processors and Cray® SupercomputersModified Apache Hadoop to report additional performance metrics, ported Hadoop to commercial Cray Supercomputer, developed an upgrade.
• Online media streaming service.Implemented Event Orchestration Layer (which was backbone of the service). It allowed very high throughput message passing throughout the system. Developed Apache Spark based module to.

Master'S Degree, Computer Science

Bachelor'S Degree, Computer Science