← Portfolio
interactive-demoPythonOpenAISQL
Retail Inventory Routing Agent
Clean-room agent demo — natural language inventory questions with expandable decision logs.
Situation
Operations teams needed faster answers about inventory rebalancing across distribution centers — ad-hoc SQL queries were too slow for iterative what-if scenarios.
Task
Prototype an agent-assisted workflow that interprets natural-language questions and returns routing recommendations using a clean-room retail scenario (no proprietary employer data).
Action
- Defined tool interfaces for inventory lookup and routing simulation
- Built API proxy layer to keep model credentials server-side
- Designed UI with expandable agent logs to surface reasoning steps
- Implemented mock mode for reliable demo performance
Result
Demonstrated end-to-end agent architecture suitable for stakeholder demos and technical interviews. Live demo available in the Lab.