At Chase
Enterprise AI & Intelligence
Firm-wide GenAI analytics platform and knowledge graph, built end to end.
40%+ query accuracy · 60% less SME dependency
The problem
Analytics across lines of business ran on inconsistent definitions. The same question asked in two teams could return two different answers, and getting a trustworthy one usually meant finding the right subject-matter expert.
What I built
I defined the product vision for a firm-wide Logical Knowledge Graph — the semantic and data-quality foundation for downstream analytics — and owned its data standards, definitions and ontology governance. On top of it, I set the requirements for a GenAI-powered query and search layer so business users get consistent, auditable answers directly.
Impact
Query accuracy improved by 40%+, dependency on SMEs dropped by 60%, and automating knowledge-graph maintenance and data quality checks cut build time by 50–70% — making the platform sustainable at scale.
Alongside it
Match Studio, a semantic RAG tool that flags similar work already in flight across LUMA, DUC, JIRA and Monday before a project starts (live in the PMT tool since August 2026); and a market intelligence stack — competitor tracking for Consumer Banking, news ingestion, and GenAI-generated company profiles for Corporate Development.