A detection engineering lab: replayable attack telemetry, Sigma rules, and regression tests that fail when a detection stops firing or starts crying wolf.
- Built a platform that validates security controls and detections against replayable attack telemetry, classifying each as verified, gap, or false positive to identify and close weaknesses.
- Applied data-exposure and access controls to an AI/automation system, including PII redaction on ingested data and access boundaries, directly relevant to securing LLM integrations and data connectors.
- Documented per-control findings and remediation recommendations across ~125 tests, producing defensible evidence for control validation and reporting.