Google Ads (Offline Conversions Infra)

• Designed and implemented data import for new data types (Custom User Params/ User Data) in the latest iteration of Google Analytics. Wrote Integration/ Load Testing, and dashboards for tracking user metrics.

• Migrated Data Import connectors to a new backend in unified infra. This service has a DAU of 7K. Added alerting for migrated customers along with Diff testing, release procedures and on-call playbook.

• Managed interns during summers to deliver various projects related to Data Import/ BQ Export.

• Designed and Implemented large-scale, distributed, low-latency pipelines using C++, Flume and Conduit Data Processing that processed 25% of total Google Analytics traffic.

• Added Reprocessing support for Bigquery Export and associated auto-validation framework. This decreased the effort required for Bigquery-only export from O(weeks) to O(hours).

In school, during the semester, I am a Teaching Assistant for CS2510: Fundamentals in Java (Introduction to Class-Based Program Design)

Responsibilities include:

1. Conducting TA Hours for students taking the course.

2. Grading answer sheets.

3. Answering questions/queries on Piazza.

• Built a complete web-app which included a database, frontend and associated backend services using C++ and Angular/TypeScript. The application is in production and being actively used.

• Reduced turnaround time for curation from around 24 hours to less than an hour by creating an alternate human assisted process.

• Developed features for naviHealth’s suite of web applications as a full-stack developer, in an agile environment.

• Designed and developed API's to communicate between applications providing low-coupling and high-cohesion.

• Improved and refactored existing services to provide better runtime performance.

• Integrated Pendo Analytics to various applications. Tagged pages and features for robust analytics collection to aid analytics team.

Worked on CUDA C to create and implement various image-processing algorithms with improved run-time by leveraging the power of parallel processing with C in the NVIDIA Architecture.