MGT 6203 - Data Analytics for Business (Summer 2024)

Built data solutions to enhance content moderation systems and ensure trust among LinkedIn users.• Review Quality: Designed & implemented a stratified sampling based framework for auditing reviewed content by human reviewers & classifiers. Used spark to engineer a reporting system for evaluation of model performance and decision accuracy, offering real time insights resulting in 3X faster reporting and 2% uplift in weekly QA scores.• Ads Trust: Led DS efforts in collabration with product & engg to successfully migrate ads review to a new platform. Created SoT UMP tables in HDFS for review funnel metrics and productivity metrics for 1000+ reviewers. Optimized ETL pipelines for metric availability, achieving over 50% improvement in runtime efficiency & precise measurement of reviewer handle time. Developed interactive Tableau dashboards to provide actionable insights.

Consulted financial institutions on treasury management solutions• Led a team of 3 & built full-stack solutions utilizing java within spring framework to develop features including automated deal reporting, confirmation & SWIFT messaging for the firm’s proprietary trading & risk management platform.• Enabled the client to arrange India’s first-ever SOFR-linked trade financing deal following the LIBOR transition in 2021.

• Modelled a random forest classifier for prediction of high risk accounts with 24.4% NCL rate to limit delinquency losses.• Devised an early credit line decrease strategy for the predicted accounts, saving an estimated $3.6M annually.