Deepanshu Jindal

Anti Money Laundering (AML)

1) Threshold Calibration: Recalibrated thresholds using Above-The-Line (ATL) and Below-The-Line (BTL) methodologies to enhance detection accuracy.
2) False Positive Suppression Model: Developed machine learning models utilizing Random Forest and XGBoost techniques to minimize false positives while maintaining minimal event loss. Achieved a 44% reduction in alerts with a 1.2% event loss. Employed GridSearchCV for hyperparameter tuning and trained the models with a focus on recall metrics. Trained and deployed 43 models in the Data Warehouse (DWH) for operational use.
3) Overlap Analysis: Identified AML rules with intersections greater than 95% either individually or in combination, enabling control over alert generation and preventing account repetition.
4) Scenario Validation: Validated the current thresholds for all existing rules to ensure the accuracy of the deployed AML application. Conducted analyses and provided observations to the client for necessary adjustments.

• Utilized advanced querying, visualization and analytics tools to analyze and process complex data sets.
• Worked with stakeholders to develop roadmaps based on impact, effort, test coordinations and Campaign Management.
• Set up ETL's and develop end to end automation for Reporting and Dashboards.
• Retail Assets (Secured Lending) : Conducted thorough data mining and analysis for end to end automation of Credit Life cycle for Housing Loan and Loan against Property including Origination, Underwriting, Disbursements, Monitoring, Repayment, Renewal/Refinancing, Default/Collection & NPA/Recovery.
• Retail Assets (Unsecured Lending) : Developed credit risk models using statistical and machine learning techniques (Weight of Evidence, Logistic Regression, Ensemble Models etc.). Conducted data preprocessing , Feature Engineering and model validation to enhance accuracy.
• Default Prediction : Built predictive models to forecast likelihood of loan defaults and Implemented early warning systems that enabled proactive risk management. Also, Leveraged Credit Bureau data extensively for holistic customer assessment.
• Credit Portfolio Analysis : Conducted Portfolio level analysis and identified key risk insights such as MOB wise Delinquency Trends (30+,60+ and 90+), Quarterly Delinquency Transitions & Rating Migration, Population Stability Index etc.
• Automation of NPA & SMA : Reporting on Daily, Monthly and Quarterly Basis which provide real-time visibility into the status of NPA & SMA accounts, enabling better & real time decision-Making by the stakeholders.
• CIBIL Triggers (Acquisition & Retention) : Envisaging immediate outreach to our existing Credit Card/Housing Loan/LAP Customers inquiring for credit facilities with our peer banks and financial institutions. Cibil TransUnion facilitates sharing data of Inquiries for Asset Customer Base subscribed on their product platform Cibil Watch. Automated the process by setting up multiple ETL's for flow of information to Contact Centre/CRM to reach out to customer in quickest time possible.
• Income Prediction/Imputation : Developed and Implemented a segmented (Secured Loan, Unsecured Loan, Standalone Credit Card and NTC Segments) model using machine learning algorithms which resulted in improved accuracy and informed data-driven business decisions such as pre-qualified and pre-eligible Loan Limits across various asset products.

• Campaign Management (HL, LAP & Gold Loan) : Leveraging Cross Sell opportunities across existing Liability and Credit Card Base by analyzing Consumer & Commercial Scrub. Furthermore, Driving the campaign across multiple Digital Channels (Mobank, Whatsaap, Email & SMS) and Automated MIS Reporting pertaining Traceability & Conversions on daily basis via Emails to concern stakeholders.
• Customer Transactional Analysis: Objective of the project was
  to build BI dashboards using SAS VIYA of Debit Card transactions
  (POS, ATM & E-Commerce) to understand customer behavior and
  help business for brand collaboration with different E- commerce.
• Home loan Propensity Model: Objective of project was to
  identify probable customers who are most likely to take home
  loans using Machine learning algorithm in SAS.
• Pre-delinquency Risk Model for Personal Loan Customers: Objective of project was to identify customers who are going to fall in 1-30 days of delinquency and hence decrease delinquency rate using machine learning algorithm in SAS.

• Gathered, arranged and corrected research data to create representative graphs and charts highlighting results for presentations.
• Validated incoming data to check information accuracy and integrity while independently locating and correcting concerns.
• DISCOM / Power Generator/Supply Chain data analysis.
• Resolving client queries.
• Maintaining & updating the Knowledge Base of Power Sector.
• Project Tracking and market research.

• Performing web based financial research required for the generation of scope-cuts using Hoovers & Capital IQ.
• Responsible for content development and its quality maintenance by Performing Auditing Checks like Reviewing the Job Matching, Cleaning the LTI based on the relevance of the data using MS Excel.
• Understanding the adequacy of the market research process with the respective stakeholders and Conducting Research required for the generation of scope-cuts.