AEM Content Validator
Mayo Clinic
- Architected a production LangGraph workflow with 5 parallel agents and a zero-cost triage router to avoid unnecessary LLM calls.
- Reduced article validation time by 94%, from roughly 2 hours to under 10 minutes per document.
- Loaded 34 validation rules from Neo4j Aura with JSON fallback for resilient, zero-downtime rule delivery.
- Built a grounded-accuracy retrieval flow on PGVector and validated faithfulness and recall with RAGAS before rollout.
- Shipped a Next.js review dashboard with SSE streaming and human approval gates using LangGraph interrupt().