Bivor Mallik

Led the development of Portfolio Risk Management, Transaction, and Payment models for Chase credit cards, while establishing an explainable AI team to focus on risk models across all retail products.

• Drove the responsible adoption of AI/ML at scale across credit card, auto, and home lending, ensuring compliance with regulatory requirements.
• Supported multiple regulatory exams (FRB, OCC, CFPB), with a key focus on the integration of AI/ML in consumer lending decisions.
• Developed and implemented a first-generation transaction-level ML model impacting 1.2 billion credit authorizations monthly, driving approximately $110 billion in sales.

Led multiple data science and modeling teams focused on Account Management risk models for Chase credit cards.

• Developed and implemented first-generation machine learning models for account management strategies impacting over 80 million Chase credit card customers, ensuring compliance with U.S. regulations regarding Adverse Action (AA) reasons.
• Created a first-generation ML model to assess the likelihood of reward gaming risk at the time of application, resulting in a reduction of over $50 million in reward costs.
• Filed three U.S. Patents related to machine learning model explainability, debiasing techniques, and actionable recourse.

Led multiple data science and modeling teams in the end-to-end development and implementation of ML and data products for Credit One Bank.

• Delivered the bank’s first-generation Acquisition Fraud Model, distinguishing between Credit Burst Out and Identity Theft, resulting in a $35M+ reduction in annual fraud losses.
• Leveraged fraud intelligence tools (MaxMind, Precise ID, Fraud Shield) to enhance fraud detection across Direct Mail and Web channels.
• Introduced Boosting algorithms in Response Modeling, improving Direct Mail campaign response rates and reducing annual mailing costs by $18M+.
• Spearheaded customer retention initiatives by developing attrition models, with a focus on high annual fee products.

Managed a team of data scientists at American Express (AXP), leading the development and implementation of machine learning solutions for Credit Limit assignment.

• Oversaw the deployment of multiple limit assignment models using advanced techniques (bagging, boosting, look-alike models), driving over $150M in profit.
• Introduced Generative Adversarial Networks (GANs) to address small sample biases, improving model performance.
• Drove team growth by expanding coverage of modeling solutions across the customer lifecycle, ensuring alignment with core business objectives.

Held multiple roles (Analyst, Assistant Manager, Manager, Senior Manager) across AXP Data Science and Risk Modeling teams in India and New York.

• Developed Acquisition models for AXP’s small business portfolio using boosting techniques and enhanced reject inference methodology, driving over $10M in profit.
• Leveraged trade line data from major consumer (Experian, Equifax, TransUnion) and commercial bureaus (D&B, SBFE) for model development.
• Contributed to successful CCAR exam outcomes by developing Probability of Default (PD), Annual Recovery (AR), and Recovery models for AXP's retail business across 6 international markets.
• Designed Collection and Recovery models for multiple markets, reducing credit losses from delinquent and charged-off accounts.