AI/ML engineering leader. Agentic systems in regulated domains — healthcare, edtech, fintech-adjacent. Director-track, 35+ years across recommendation systems, enterprise AI/ML delivery, and education-technology curriculum at scale.
Two foundational patents — US 6,850,988 (e-commerce personalization, foundational to Amazon's recommendation engine) and US 6,839,229 (large-grained database concurrency for data warehouses). Formerly Master Technologist at Hewlett-Packard and Principal TPM-AI at Microsoft.
- pacca — Multi-agent AI platform that automates healthcare prior-authorization decisions. RAG-grounded, harness-engineered, observability-first. 85–90% workflow time savings on synthesized cases.
- adult-learning-coach — Agentic coaching system that turns distance-learning recordings into evidence-based instructor coaching reports. 20–25% time savings.
- lesson-planning-assistant — K-12 lesson planning agent built on the Claude API with a 4-pass refinement workflow.
- 📄 CRISP-AG: An Artifact-Centered Framework for Enterprise Agentic AI Governance (v2.3, May 2026) — practitioner white paper specifying the four implementation artifacts under-defined by ISO/IEC 42001 and NIST AI RMF.
- 📚 Specification-Driven Design + Consolidated PRD for PACCA — companion docs walking the v2.3 harness-engineering cycle from problem statement to deployment.
- 🎓 Agentic AI Implementation Guide for K-12 Education — framework for training teachers on AI management competencies.
The full curated catalog of code, papers, PRDs, patents, and career guidance lives at drdavidreed.com/portfolio.
PACCA v2.3 harness-engineering cycle — adapting Lin et al.'s Agentic Harness Engineering methodology to a healthcare-governance domain, every behavioral change a one-file diff with a falsifiable predicted-impact contract.