Priyanka Bhutoria

Built a Boosting-based propensity score model for Advanced checkout product, yielding a 5% uplift in marketing campaigns, based on data from around 10 million US merchants. Worked on anomaly detection model to identify bot attacks on login portals, reducing security risks and enhancing user experience. Conducted analysis, comparing various hypotheses (Bollinger Bands, Z-score, KPI trends) to identify the best combination. Analyzed existing data to identify PayPal users on merchant websites, improving Shopper Insights and Personalized Offers. Conducted research paper reviews related to this use case.

Developed a system to strategically reduce expenses associated with Blue/Green deployments. Designed and executed AWS Glue Spark job, resulting in a 98% improvement in job time. Implemented a streamlined data storage system in AWS S3 and OpenSearch, yielding critical insights for prioritizing subsequent projects aimed at migrating properties from B/G to in-place. Enabled enhanced data utilization by making S3-stored result data available for advanced analysis utilizing Window functions in AWS Athena. Contributed to AWS OpenSearch Hackathon, analyzing domain change time distribution for improving processes and customer service.

Conducted an in-depth analysis and evaluation of diverse NLP models aimed at optimizing the next word prediction model. GRU model gave the best result with 82% accuracy.