Responsible for the monitoring and validation of a suite of 42 US MarketModels (including Risk, Propensity, Fraud, Mortgage models).Taking action on client requests involving RCA, comparative analysis,benchmarking of models.Acting as a voting member and reviewer of model design, model concept andmodel documentation for the upcoming models.Maintenance and upgradation of Model Risk Index.Working on Implementation of a standard Model Governance and Validationprocess across all regions around the globe.Acting as the secretary of the Global Model Governance Board.

Worked on automating the model validation system using R, to generate finalR-markdown reports for models.Developed a framework to risk rank all the models in inventory based on there risk factors.

Propensity Model: Designed a modeling methodology for predicting customers'propensity of responding to card balance transfer offers.Amount Model: Successfully predicted the amount of debt on credit cards acustomer would transfer to a Citi credit card, thus enabling accurate targetingof more profitable customers.Model Performance Analysis: Designed an analysis to understand theperformance boost and increment in revenue brought by the new generationmodels compared to old generation models.The new generation Propensity and Amount models are estimated to driveUSD 100 MM incremental revenue over 3-5 years.Other Responsibilities: Monitoring the performance of existing models.Providing data and information for model Audits.Algorithms & techniques used : XGB, Logistic Regression, Linear Regression.

Developed model to predict the booking amount by recognizing discounting pattern through distribution with an accuracy of 98% (MAPE).Predicted the expected number of days to close an open job opportunity in Red Hat with MAE value of 21 daysWorked on forecasting quarterly bookings using time series models.Modelled data to predict the booking amount of run rate business with 85% accuracy.