Current - ML Platform for GPay

– Tech Lead on the GPay Transaction Intelligence ML team.

– Led the development and launch of Embedding service which generates consumer / merchant embeddings for 370M users based on

GPay India transaction data.

– Improved downstream voucher models integrated with user embeddings led to 3% increase in voucher revenue and 7% increase in user

engagement.

– Designed, developed and launched a deep learning model to generate a confidence score for GPay India consumer / merchant address

accuracy.

– Address score is used to improve overall address quality, improve targeting of users / merchants for credit products, and improve credit

lender efficiency by reducing contact point verification (CPV) before disbursal of credit.

– Developed and launched GPay India GenAI based user employment categorization by training a model using embeddings and transaction

description history as features. Gemini inference was used to generate the training dataset.

– Leading the effort to develop a Gemini GPay extension which provides a seamless Gemini interface for any queries requiring access to

user’s personal data.

Past - Ads Incentives

– TL for the Sales Incentives Platform which powers incentive programs across Google and YouTube Ads.

– Led migration of incentives platform from Payments to Ads without any downtime.

– The migration involved a complex validation of new infrastructure by replaying every request received on the existing system while running

both systems in parallel.

– Led design and development of Acquisition Incentives Fund (AIF) program targeted towards SMB advertisers.

– AIF led to onboarding of 500 new advertisers / agencies to Google and drove $200M in tail acquisition revenue with a bonus payout

of $50M to partner agencies annually.

– Collaborated with stakeholders and x-fn teams ranging from sales, marketing and legal to launch new incentive programs on the platform.

Azure Data Lake Store and Cosmos Store

– Designed and developed the event notifications feature for ADLS (third-party) and Cosmos (first-party) which allows applications to

react to certain file system operations.

– Created an event driven pipeline using serverless architecture that responds to changes in data as they occur.

– The events are stored in an in-memory durable metadata store based on RSL, integrated with Azure Event Grid and are delivered with

At-Least-Once semantics.

– Developed a framework to enable cross-account event notification delivery system using shares.

– Implemented data share feature to enable cross-account shares in ADLS and Cosmos based on Access Control Lists.

– Designed and implemented transformation of file system enumerates, increasing performance and throughput by 3x.

– Designed and implemented an auto-tracing framework which is responsible for providing audit logs to customers.

Yahoo Mail Storage team

– Designed and implemented a prototype for cross-cluster replication of yahoo mail messages using Apache Kafka.

– Implemented RESTful Web Services in Java for asynchronous replication in order to reduce latency.

– Built test suites for testing using TestNG and Mockito unit testing frameworks.

Digital Printing Group

– Involved in application software development for Multi-Function Peripherals(MFPs).

– Handled the network modules of MFPs running on RTOS, including features like Scan to Email, Shared Folder, Scan to SMB, User

Authentication etc. which comprise network protocols like SMB, SMTP, SMIME, LDAP etc.