AI governance · Operational readiness · Workflow design
My work sits between policy and execution: understand how the work actually moves, decide where automation belongs, make authority and evidence explicit, and test the design before it becomes a production problem.
Live application
A referral workflow that helps employees act on open roles without creating a second candidate database. I defined the workflow, data boundary, ATS system-of-record rule, review behavior, and trust requirements; implementation was AI-assisted.
Research archive
Case studies of failures where authority, feedback, safeguards, or operator understanding broke down. Each entry asks what control was missing and what the failure teaches about AI-enabled systems.
Governance contribution
Contributed to operationalizing a pre-deployment assessment that asks whether an AI use is necessary and justified before implementation, rather than starting from the assumption that AI should be used.
Start with the work as it happens: objective, actors, handoffs, data, systems of record, decision rights, incentives, and the places a failure can pass unnoticed.
Define what needs evidence, who can approve what, where exceptions go, when work must stop, and what changes require another review.
Use an AI-assisted prototype, workflow model, or tabletop scenario to expose assumptions and failure paths before treating the design as ready for production.