Senior Manager

Ernst and Young

1. Project 1: Model Validation & Portfolio Segmentation
3. Role: Responsible for model validation across various portfolios developed with XGBoost.
4. Responsibilities:
5. Validated model performance using KS statistic & Decile wise Waterfall.
6. Computed PSI to understand current data scoring compared to model development data.
7. Calculated CSI to analyze distribution differences in explanatory variables.
8. Published monthly model monitoring reports for governance.
9. Conducted portfolio segmentation to enhance customer onboarding experience.
10. Developed an Income Estimation Model for rural customers lacking bank statements.
11. Project 2: Risk Mitigation & Business Growth
13. Role: Led initiatives to mitigate risks and enhance profitability using data science.
14. Responsibilities:
15. Developed a machine learning model for Early Warning Indicators.
16. Analyzed collections data to improve targeted campaigns.
17. Utilized Cat Boost algorithm to predict default risks.
18. Created Acquisition Scorecard for onboarding new Business Loan customers.
19. Conducted exploratory data analysis (EDA) for insights and patterns.
20. Designed Behavioural Scorecard for customer onboarding and cross-selling ev

Lead Data Scientist

HCl

1. Roles and Responsibilities:
3. Risk Consulting and Business Growth Strategy:
5. Conducted reject inference analysis and consulting for top banks in India to identify business growth and risk management opportunities.
6. Executed swap in/swap out analysis to enhance consumer portfolio and SME strategies.
7. Renewal Project Leadership:
9. Led a renewal project resulting in high Return on Assets (ROA) and profitability margins.
10. Achieved significant profitability by halving the cost of default in renewal cases and reducing the Cost of Customer Acquisition (COCA) to zero.
11. Fraud Detection:
13. Developed Fraud Score models using XGBoost and Random Forest algorithms to predict first-time fraud transactions.
14. Implemented classification models to identify fraudulent behavior among healthcare providers.
15. Microfinance and Subprime Market Analysis:
17. Designed Microfinance scorecards using logistic regression for fintech platforms like SME-Lending Kart and Paytm.
18. Built solvent score models using Random Forest to identify creditworthy customers in the subprime market.
19. Behavioral Analysis and Scorecards:
21. Utilized Weight of Evidence (WOE) and Information Value (IV) tools to predict customer behaviors.
22. Developed behavior scorecards for renewal, top-up, and early warning systems for collections and recovery prioritization.
23. Lead Prioritization and Recommendation Engine:
25. Created lead prioritization frameworks using ready reckoner methodologies.
26. Designed and implemented real-time recommendation engines to rank sales leads for upsell opportunities.
27. Cloud Migration Experience:
29. Successfully migrated on-site solutions to cloud-based platforms such as AWS (EC2), ensurin

Data Scientist

IBM

1. Project 1: Personal Loan Model-Credit Risk Application Score Card Model.
2. Responsibilities:
3. • Worked with respect to customer behaviour and predicted the reason for rejection using SHAP
4. values for each customer.
5. • This enabled the client to give loan to only non-defaulter.
6. • Performed data cleansing, data manipulation and exploratory data analysis to identify, analyse and
7. interpret data to find out trends/patterns in large data sets.
8. • Predicted behaviours with tools like Weight of Evidence (WOE) and Information Value (IV).
9. • Worked with large data sets and cleaned the data with fine classing and coarse classing.
10. • Alternate Data helped to handle NTC cases.
11. • Defined good definition, bad definition.
12. • Reject Inferencing will help us to tag the performance for rejected customer.
13. • Building model with Reject Inferencing will make our model less conservative and more realistic.
14. • Build solvent score (Random Forest) model to find good score.
21. Project 2: Built underwriting model to find good credit customers in subprime market.
22. Responsibilities:
23. • Performed data cleansing, data manipulation and exploratory data analysis to identify, analyse and
24. interpret data to find out trends/patterns in large data sets.
25. • Predicted the behaviours with tools like Weight of Evidence (WOE) and Information Value (IV).
26. • Worked with large data sets and cleaned the data with fine classing and coarse classing.
27. • Built Micro finance scorecard (Logistic regression) for Indian Microfinance market.
28. • Build solvent score (Random Forest) to find good credit customers in subprime market.
29. • Alternate data helped to give holistic picture of customers to get better prediction.

Data Analyst

Accenture

1. Project 1: To classify Fraudulent behaviour of healthcare providers.
2. Responsibilities:
3. • Technical review of the user stories scoped for the entire release.
4. • Technical implementation and code reviews.
5. • Analyse the complex business needs of clients and convert them too functional and technical
6. specifications.
7. • Responsible for solution design feasibility, technical design, development, testing.
8. • Worked on modelling projects involving model base data preparation, customer info encryption/
9. decryption process, data cleaning, bivariate analysis, pairwise correlations, outliers, logistic
10. regression techniques.
11. • Model development using logistics regression, linear regression
12. • Strong exposure to data preparation for model development (application & performance data)
13. • Exposure to trees in python (Random Forest)

University of Texax and Austin

Post Graduation in AI and ML