Model Risk Management App
Revolutionizing Lending Risk Management: Our team revamped a legacy system, harnessing modern technologies like machine learning to provide real-time data on top-performing lending profiles, automating notifications, and ensuring efficient lending risk management.
.png)
Nov 2020 - Dec 2021
TLDR
Banks face inherent risks in lending, but modern technologies like machine learning have transformed the industry. By leveraging key variables, algorithms can assess lending risk and generate new profiles. However, monitoring and maintaining model performance pose challenges. Our team revamped a legacy system, modernizing it to provide real-time data on top-performing lending profiles. We automated notifications and tracked model performance, ensuring efficient management of lending risk.
My Role
As the pioneering UX/UI Designer on the Minerva app project, I spearheaded the design process, from gathering requirements to establishing design styles. Responsibilities included architecting user flows, collaborating with stakeholders to craft a design roadmap, creating wireframes and prototypes, and overseeing design implementation and team onboarding. I also created the first atomic design system for the product.
What
The Minerva app addresses the need for Wells Fargo to manage lending risk in a B2B context, adhering to strict FDIC governance. The legacy system lacked the capacity to track and manage model risks effectively, hindering intuitive review and approval processes.
Why
With statistical models crucial for assessing loan applications, it became imperative to enhance model management capabilities. The solution aimed to streamline model oversight and ensure compliance with regulatory standards.
How
After consulting with stakeholders through a series of interviews and focus groups, it was evident that a new application was necessary to manage models effectively. Key features such as timelining, tracking, notifying, and scheduling were identified to maintain model relevance and monitoring. Despite challenges in determining notification protocols for various user groups, we devised adaptable workflows to accommodate evolving user needs.
What we learned
​Comprehensive task analysis and user flow mapping, helped us in the early days of discovery as we established a user-centric design approach. Studying each user's journey during the model approval process identified crucial touchpoints and optimal notification moments for reviewers. The introduction of new features necessitated the creation of new application sections and enhancements to existing ones. Utilizing low-fidelity designs proved instrumental, allowing for agile iterations and effective stakeholder presentations without getting bogged down in details.