Kaiser F. Email & Phone Number

As a principal data scientist at Abrigo, I lead the data science practice in developing and deploying AI/ML products aligned with the company's risk and growth offerings, and in defining the structures needed for ongoing AI/ML work.

Mclean, VA

• As a principal data scientist in Capital One's Small Business Banking team, I developed, deployed, and managed models to assess credit risk for lending decisions, and fulfilled audit requests to keep model design and.
• Developed ensemble GBM and logistic regression model in Python and deployed on AWS, drawing on internal lending and deposits data, and external vendor data, enabling daily risk rating and monitoring of entire small.
• Fulfilled audit requests in record time by building models and code as audit-ready and documented from inception, tracking all code development and auditor questions through GitHub, and responding to all auditor.
• Built automated Python modules to explore data and leveraged statistics knowledge to identify lending risk drivers, enabling real-time communication of individual risk scores to underwriters and analysts.
• Redesigned monitoring scripts for loan originations risk model by joining quarterly and weekly processes, automating data loads, summarizing risk features, loading results into a navigable dashboard, and revising.
• Designed and implemented manual override process to enable underwriters to override automatic risk ratings for specific customers, and to track the reason and approval process for overrides, ensuring process was.

Washington, District Of Columbia, United States

• As part of IHS Markit's Advanced Analytics team, I developed statistical models and algorithms to support automotive, chemicals, energy, and maritime teams. Key accomplishments:
• Led team of 10 data scientists across 2 countries through exploration of automotive data with over 500 million purchase records and design of forecast models.
• Developed ensemble logistic regression and decision tree model to predict timing and types of future vehicle purchases.
• Developed tool to forecast demand for 199 chemicals in 88 countries, replacing 6-month cycle of manual adjustments to 200 Excel files with a 1-day automated process in R ingesting data from AWS Redshift.
• Built methods in R to measure impact of qualitative factors on crude oil prices using model residual analysis, quantifying the effect of OPEC meetings and geopolitical events.
• Streamlined health care simulation model of supply and demand of health care services and workers, enabling quality checks and reducing calculation times from running overnight to under 1 hour.

Washington, District Of Columbia, United States

• As part of IHS's consulting services, I managed contracts with government clients to optimize data processes. Key accomplishments:
• Managed 5 consecutive year-long contracts up to $2 million in value, leading teams of 4-10 developers through successive code optimization projects.
• Modernized publication process for over 900 tables and reports, replacing over 100 processes requiring 40 staff to maintain with 6 processes in SAS Enterprise Guide requiring fewer than 10 staff to maintain.
• Designed modernized processes for data management and ad hoc reports using SAS Enterprise Guide and SAS Business Intelligence applications.
• Drafted over 200 pages of user documentation and led 4 week-long training sessions to transfer ownership of modernized process to client.
• Developed and implemented k-nearest neighbors' algorithm to classify workers' compensation recipients based on nature and type of injury.

Washington, District Of Columbia, United States

• As part of IHS's consulting services, I developed and deployed statistical models to guide decision-making in disability review processes. Key accomplishments:
• Built text-mining process in SAS Enterprise Miner to clean 20 million freeform text responses with 100,000 unique medical terms, cluster results, and build a predictive logistic regression model.
• Redesigned logistic regression models in SAS used to guide decision-making in type and frequency of review, resulting in saving up to $20 million in annual review costs.
• Drafted over 140 pages of user documentation to enable client to understand results and take ownership of programs.

Master'S, Public Policy

Bachelor Of Arts (B.A.), International Economics

Bachelor Of Arts (B.A.), Economics