Harmful Information · Detection, Triage & Policy
I monitor harmful information in humanitarian emergencies — and build the systems that keep that monitoring honest.
I'm a Senior Officer for Communications Monitoring and Analysis at the International Federation of Red Cross and Red Crescent Societies in Geneva. My work sits where information risk meets humanitarian operations: impersonation and donation-scam campaigns during disasters, health disinformation during outbreaks, narratives that put field staff at risk.
When the tooling I needed didn't exist, I built it — three autonomous detection-and-triage pipelines on the Claude API, running daily through a live public-health emergency. I also work the institutional side of the same problem: co-drafting the Red Cross Red Crescent Movement's 2026 resolution on harmful information, and supporting the launch of its joint Movement Hub on Harmful Information.
Keeping an LLM pipeline honest during a live public-health emergency
Three autonomous harmful-information detection-and-triage pipelines on the Claude API, run daily through the 2026 Ebola PHEIC.
Read the case study → System designAn executive-readout architecture for monitoring digests
A config-driven, multi-client monitoring engine that leads with judgment — tier assignment decided in code, never by the model.
Read the design → R&D · in developmentMirrorline — auditing what AI assistants say about an organization
Structured probes to AI assistants, audited against the public record. The divergence is the finding.
Read the approach →