Data Tech 2026

In the current dynamic financial landscape, effective credit risk monitoring necessitates a harmony between advanced data science techniques and human expertise. This presentation explores the development of a comprehensive credit risk monitoring system that integrates complex data pipelines alongside domain experts from finance and investment management. We will address the challenges of harmonizing multiple data sources through both internal system changes and external services. We will discuss various modeling approaches that were considered including the Hidden Markov Models and XGBoost.

As a centralized data science team, domain experts from the business team are crucial to our development process. This collaboration ensures that financial nuances and industry-specific features are effectively captured and explained, enhancing predictive capabilities and transparency for end-users. We will discuss considerations for model delivery as well as a measurement framework to evaluate model benefits across a multi-stage production plan to establish trust while advocating for data science benefits. Attendees will gain insights into how this integrated framework not only improves risk prediction but also fosters a dynamic collaborative environment for continuous innovation and learning.