GenAI Risk Intelligence Tool

Overview

Developed a generative AI document classifier to auto-extract credit risk signals from structured and unstructured loan documentation, paired with an LLM prompt evaluation interface for non-technical analysts.

The Challenge

Risk analysts spent hours manually reviewing loan documentation, and existing tools lacked the ability to process unstructured data or explain AI decisions to non-technical stakeholders.

The Approach

Partnered with data science and legal to develop a generative AI classifier for loan docs. Built an internal prompt evaluation tool and explainability dashboard to surface model fairness gaps and drift indicators.

Key Outcomes

• Built GenAI document classifier for auto-extracting credit risk signals
• Created LLM prompt evaluation interface for non-technical risk analysts
• Developed AI Ethics Flagging Dashboard surfacing fairness gaps and drift indicators
• Enabled compliance teams to test and refine prompts on OpenAI/GPT APIs

The Result

Non-technical analysts gained self-serve access to LLM-powered risk analysis, reducing review cycles and ensuring regulatory compliance through built-in explainability features.