Synthetic Data Generation for Credit Risk using GAN Models | R&D Project- Developed GAN models to generate synthetic data, addressing class imbalance in delinquency datasets.- Impact: Improved credit risk model performance by integrating synthetic samples with real data, enhancing predictive accuracy and risk assessment.

Early Identification of High-Risk Cohort Customers | European Telecom- Led the development of custom Behaviour Scorecards for a leading telecom client in EMEA by utilizing historical payment patterns, bureau data and microeconomic factors to identify high-risk customer cohorts.- Impact: Early detection of high-risk customers allowed the client to save additional 14Mn Euro annuallyTwo Stage Expected Collections Recovery Models | EMEA Bank- Developed two-stage late collections recovery models to predict the propensity of delayed payments and identify customer cohorts to estimate recovery.- Impact: Bank was able to make recoveries for ‘most likely to repay’ customer segment with 73% accuracy

Store Prioritization for HVAC Maintenance | Canadian Retail Chain- Build a machine learning model to incorporate multiple internal and external factors to classify and prioritize retail stores for HVAC maintenance, minimizing downtime and revenue loss.Headroom Analysis to Identify Potential Subscribers for a New Drug | Fortune 500 Big Pharma- Developed a predictive model to identify medical professionals with high propensity to prescribe the drug, based on pharma representatives' data, prescription history, insurance coverage, and market dynamics.- Improved targeting strategies for medical representatives, leading to better allocation of resources and increased market penetration for the drug.

- Understanding the fundamental designs of heavy electrical machinery- Understanding the process of heavy machinery and power systems fabrication