AI that ships
AI products & agentic systems
LLM features people actually use: architecture, rollout, evaluation, and guardrails, not demos.
Selected experience 3
- Set technical direction and led end-to-end UI architecture for the Meta AI Assistant on Quest: 42% engagement lift within 30 days of launch, then a staged 800K → 8M+ user rollout with telemetry-driven rollback.
- Build LLM-driven agent workflows at Marlow that draft and file CMS-compliant nursing documentation into PointClickCare, with tool-use integration, prompt and output evaluation, and validation guardrails.
- Earlier ML product work: heuristics-to-ML transition for Suggested-As-You-Type across Messenger clients, and NLP entity extraction over 100K+ agreements at Bloomberg.