Retrieve incident context
Domain-specific RAG
Built retrieval over historical incident information to bring relevant prior events into root-cause analysis.
Retrieval and agent workflows that bring incident history into investigation and triage.
A public sketch of the documented components and workflow. Internal interfaces are omitted.
Download full-resolution SVGIncident investigation depends on context scattered across historical records. Teams need to find related events, identify patterns and form useful root-cause hypotheses.
Domain-specific RAG
Built retrieval over historical incident information to bring relevant prior events into root-cause analysis.
LangChain · CrewAI
Engineered incident-management agents and orchestration for investigation and alert prioritization.
Historical patterns · root-cause hypotheses
Connected incident history with agent workflows to identify patterns and produce context for triage and root-cause investigation.
Selected work from my engineering role at ServiceNow. Impact figures are reported from this work; the companion examples below use synthetic data.
A new, runnable example that scores saved predictions against expected outputs. It uses fictional data and illustrates evaluation mechanics; it does not reproduce the production system or substantiate the impact figures above.
python3 evaluate_outputs.py incident-triage.json report.jsonPython 3.10+ · Standard library only · Includes fixtures, metric definitions and tests