Hands-on reference implementation · Azure AI Search
Enterprise Helpdesk Knowledge Search
A searchable incident-knowledge experience that retrieves related issues, previous root causes and proven resolutions from historical IT support incidents.
The problem
What this project is solving
Support engineers often search old tickets manually even when the organization has already solved a similar problem. The goal was to structure incident history so a short symptom phrase can surface relevant operational knowledge quickly.
Implementation
What I built
- Created a dedicated Azure AI Search index for incident ID, service, severity, issue, root cause, resolution and status.
- Loaded 12 synthetic incidents spanning 12 Azure services through Python.
- Tested search behaviour in Azure Search Explorer and tuned practical short-phrase queries.
- Built a Streamlit interface with service and severity filters, root cause and previous resolution.
Reference implementation using synthetic incident data; not presented as a production client deployment.
Evidence
Reference results
Delivery perspective
Why this matters
The project demonstrates a simple but useful delivery pattern: model operational knowledge correctly, make retrieval easy for the support engineer, test relevance and expose the output in a workflow-oriented interface.