Divyansh Gupta

PREDICTIVE FRAUD INTELLIGENCE (PFI)

• Led feature engineering initiatives to build machine learning models for scoring and predicting fraudulent transactions, serving major clients such as Coelsa, New Pay, BaseLine Model, and Pix.
• Redesigned the feature engineering pipeline, achieving a 90% improvement in feature delivery performance, which significantly accelerated model training, and positively impacted product timelines.
• Developed an automated framework for feature creation, eliminating the need for individual code writing, thus enhancing efficiency and scalability.
• Collaborated with data scientists to understand and implement new feature requirements, ensuring alignment with business objectives.
• Designed and implemented a feature correctness and testing framework, establishing rules for each feature family to maintain data accuracy and integrity.
• Improved developer experience, code quality, and modular design through the redesign of key modules, promoting better maintainability, and scalability.
• Attending daily stand-up meetings to update the onsite and offshore teams.
• I received the Visa Innovation Award for building the Feature Explainability tool to showcase the lifecycle of a feature in the A2A feature store.
• Collaborating with cross-functional teams to productionize A2A fraud detection models.

DWH

• Develop and maintain multiple data pipelines to ingest raw OLTP data coming from various sources to S3, and create Delta tables on top of it in order to build data cuts and data marts for business analytics, enabling a lakehouse architecture.
• Developing PySpark jobs based on the business rules provided by the PO team to provide DAAP within a lakehouse architecture using Delta Lake.
• Debugging and providing BAU support on an iteration basis.

GLOBAL COMMERCIAL PAYMENTS

• Reading SORs of different entities from HDFS and receiving files through SFTP from the product and business team, after that massaging the data using PySpark and creating DAAP (facts and dimensions) to be utilized by the business team for reporting needs.
• Developing PySpark jobs based on the business rules provided by the PO team to provide DAAP.
• Upgrading the existing BCV system from SAS to the new Big Data (HDFS) that is further used to incentivize Amex sales personnel.
• Created key KPIs, like charge volume for consumer, SBS, and CORP cards, for enterprise.
• Designing a data warehouse schema for payments, GCS to outsource data to various consumers for their reporting needs within a lakehouse architecture.
• Implementing business rules in the data to get insights.
• Developing SQL queries to perform ETL workflows.
• Debugging and providing BAU support whenever required in the PI.
• Orchestrating a data pipeline using AMEx's in-house tool, Event Engine, as well as Java Bean workflows.
• Attending daily stand-up meetings to update the onsite and offshore teams.
• Migrating jobs from Magellan to Spark for better optimization.
• Developed utility to understand the fill rate and data quality of the SORs and dimensions.

DATA MODERNIZATION.

• Developed PySpark jobs for data ingestion from DB2, creating dimensions and facts in Snowflake, enabling a Lakehouse architecture for unified storage, processing,and analytics
• Implemented business rules for data insights and analytics.
• Created SQL queries to execute ETL workflows efficiently.
• Utilized SNS notifications for job failure alerts and monitoring.
• Deployed framework on Kubernetes cluster to enhance scalability.
• Participated in daily stand-up meetings to synchronize onsite and offshore teams.
• Authored BDDs for comprehensive testing of code functionality.

• Developed the Python-based project that reads Informatica mapping and converts its equivalent Talend mapping.