Back to projects
Jul 01, 2026
2 min read

Inventory Optimization (Human-in-loop AI)

Built a simulation-first, AI-assisted decision-support workflow for diagnosing over-range inventory and helping controllers turn SAP evidence into governed interventions.

Context. Inventory controllers needed a clearer way to understand why production orders created over-range inventory and which cases were worth acting on first.

Approach. Designed a simulation-first Python pipeline that combines synthetic detection data with simulated SAP evidence, performs deterministic root-cause diagnosis, and produces both a controller worklist and audit-oriented records. Added an AI investigation layer for grounded case exploration, alongside a planner workbench for portfolio prioritization, evidence review, recommendations, human decisions, and follow-up.

Impact. Created an auditable foundation for reducing over-range inventory while keeping planners in control. The pilot separates one-time order corrections from recurring master-data issues and makes unsupported production justifications visible for review.