| • |
Provided statistical consulting on the development
of initial artificial neural network systems for real-time
detection of credit card fraud for a leading fraud and credit
risk modeling software firm. |
• |
Developed state-of-the-art statistical models to account
for default correlation for underwriting credit insurance;
models became the standard tools for the country’s
largest credit insurance firm. |
| • |
Served as consulting expert and analyzed data for a major
credit card transaction processor. Managed development and
implementation of artificial neural network credit card
fraud model and created strategies to integrate results
into business systems. Expanded scope of original product
to include capabilities for strategic pricing and account
management campaigns. |
• |
Developed and managed the marketing and analysis department
for a major auto finance company. Created four new direct
marketing programs, including the introduction of direct-to-consumer
lending products. Aligned price and risk-on-product tiers
to achieve above-hurdle return on investment (ROI) performance
on all products and customer segments. Constructed target
population segmentation algorithms to maximize ROI and net
present value (NPV) of marketing expenditures. Built and
instituted credit risk models to lower risk while raising
approval rate. Marketing and product improvements resulted
in 100 percent growth rates while reducing credit risk exposure.
The business line currently contributes a significant portion
of the firm’s non-credit card net income. |
| • |
Analyzed expected credit loss improvement for a finance company
that was adopting a custom-built Fair Isaac credit-scoring model for
underwriting subprime auto finance contracts. |
| • |
Provided expert consulting services as part of arbitration
proceedings involving a dispute over residual value insurance coverage
and reinsurance in connection with portfolio of automobile leases.
Constructed a stratified sampling strategy to evaluate various coverage
issues and audit key underwriting characteristics for the selected sample.
Analysis included critiques of statistical sampling techniques used by other
experts involved in the proceedings. |
