Enterprise Finance · Retail & supply chain
Corporate-Card Expense Classification
AI Solution Architect & Lead AI Engineer · H-E-B
A LangGraph agentic workflow that replaced manual expense coding on regulated PII data and discovered a governed taxonomy along the way.
- classified at high confidence
- 99.76%
- transactions
- 51,518
- resolved in a single LLM pass
- 98.4%
- governed categories discovered
- 44
The problem
Corporate-card transactions were coded by hand against an inconsistent category list, on data containing regulated personal information.
What I built and led
- Architected and led a LangGraph agentic workflow on Vertex AI that discovered a governed 44-category taxonomy and then classified every transaction against it.
- Designed for regulated data from day one: governed categories, confidence scoring, and automatic OpenAI-to-Gemini failover so the pipeline keeps running when a provider does not.
Results
- 99.76% classified at high confidence
- 51,518 transactions
- 98.4% resolved in a single LLM pass
- 44 governed categories discovered