Jan 2023 – Aug 2024 · 1 yr 7 mos · Bengaluru, Karnataka, India · Remote

• Fraud Model:
• I developed a highly effective Fraud model aimed at detecting Straight Rollers within the Personal loan portfolio. By utilizing advanced data analytics and machine learning techniques, I successfully identified and prevented fraudulent loan applications. This model significantly enhanced the overall security and integrity of the loan application process, protecting the organization from potential financial losses.
• New to Credit Model:
• In my role, I created the NTC (Paytm Thin Bureau Thin) Acquisition Model for Personal loans. This innovative model optimized customer acquisition strategies by targeting individuals with limited or no credit history. By leveraging alternative data sources and advanced analytics, I enabled the organization to make informed lending decisions and streamline the loan application process for individuals who were new to credit. This resulted in increased loan approval rates and improved customer satisfaction.
• Acquisition Model:
• As part of my responsibilities, I developed an Application Scorecard specifically designed for evaluating Personal loan applications. This scorecard incorporated various factors related to creditworthiness and risk assessment, enabling accurate and efficient evaluation of loan applications. By implementing this model, the organization was able to make informed decisions based on data-driven insights, resulting in reduced default rates and improved portfolio performance.
• Behavioral Scorecard :
• Spearheaded the creation of a groundbreaking personal loan payments feature, revolutionizing the mapping of customer payment behavior. Achieved an impressive 88% AUC in behavioral scores by leveraging Paytm features and the newly developed payment feature. Conducted comprehensive vintage and roll rate analysis to define optimal target parameters. Engineered efficient data and scoring pipelines, ensuring seamless integration and operational excellence.

Jul 2021 – Dec 2022 · 1 yr 5 mos · San Jose, California, United States (Remote)

• 1. Application Scorecard:
• As a Data Scientist, I spearheaded the development of state-of-the-art Credit Risk Scorecard Models for over 32 clients in the autoloan industry. I utilized statistical predictive models like logistic regression to create tailored scorecards that accurately assessed credit risk. Additionally, I improved the existing scorecard models for older clients by incorporating additional features and leveraging advanced data analytics techniques. These scorecard models enabled our clients to make informed lending decisions, resulting in improved portfolio performance and reduced default rates.
• 2. Behavioral Scorecard:
• In this project, I focused on assessing the risk of existing customers based on their recent accounting transactions, financial information, repayment performance, delinquencies, credit bureau data, and overall relationship with the lender. I employed a combination of intensive manual feature engineering and utilized the open-source Python library Featuretools for feature abstraction and selection. By leveraging the power of machine learning techniques such as XGBoost and utilizing Optuna for hyper-parameter tuning, I developed a robust Behavioral Scorecard model. This model accurately predicted the riskiness of clients, enabling the lender to proactively take preventive actions and minimize potential losses.
• 3. Optimize Cash Collection Using Machine Learning to Predict EMI Payment Delay:
• In this project, I applied machine learning models to predict the likelihood of EMI payment delays. By analyzing historical cleared invoices, I developed a Payment Predictions Business Content that could accurately forecast when outstanding invoices for EMI would be paid. By identifying invoices that were likely to be paid late, I enabled collection managers to focus their efforts on customers with high overdue receivables. This optimization of collection strategies led to improved cash flow and enhanced efficiency.

Jul 2019 – Jul 2021 · 2 yrs · banglore

• Implemented Deep learning, Reinforcement Learning, recommender systems for Database performance optimization solutions at a production scale.
• Deployed auto-scalable machine learning application on Google cloud platform with Kubernetes.
• Organized organization-wide Tech Meetup on Artificial Intelligence.
• Participated in organization-wide Machine Learning and Cloud-Native Hackathon.

Jul 2017 – Jul 2019 · 2 yrs · Greater Delhi Area

Aug 2016 – Jul 2017 · 11 mos · Delhi, India