AIOps
NotebookLM for SRE: build a source-backed incident dossier
Use a curated notebook to compare runbooks and incident evidence, while keeping citations, freshness, and operational authority in view.
Guides
Walkthroughs for observability, incident response, and AI-assisted operations.
AIOps fundamentals
Start with what AI can do for operations, and where judgment still matters.
Read AIOps fundamentals ↗Observability for SRE
Make metrics, logs, and traces work together.
Read Observability for SRE ↗Incident management with AI
Build an incident response practice that learns.
Read Incident management with AI ↗Practical articles
Instructions, examples, and implementation notes.
AIOps
Connect PagerDuty, Jira, and Slack without losing incident state
A design guide for an incident assistant that handles duplicate events, partial failures, and reviewed AI summaries across tools.
How-To
Canary deployments: limit exposure and test the comparison
Choose representative traffic, define promotion criteria, and verify rollback compatibility before expanding a release.
How-To
Mean time to detect: calculate MTTD and expose measurement gaps
Use consistent onset and discovery timestamps, include customer-reported incidents, and interpret the average with its sample and uncertainty.
How-To
Metrics for observability: choose the question before the chart
Counters, gauges, and distributions answer different questions. Build metrics that preserve scope, denominators, and the evidence behind an incident.
AIOps
Python logging for operations: record failure and preserve the exit status
Use Python logging to capture useful context without swallowing errors, duplicating exceptions, or exposing sensitive data.
AIOps
Collect and aggregate operational data with Python
Validate the source data, preserve useful denominators, and distinguish a failed collection from a real zero before aggregating metrics.
How-To
Observability logs: preserve the events an investigation needs
Design logs around useful events, searchable context, and a collection path whose failures are visible.
How-To
Distributed tracing: choose the first request path to instrument
Start tracing with a customer-critical path, then test propagation, sampling, and whether the trace supports a real investigation.
How-To
Read a distributed trace without mistaking it for the whole system
Understand spans, parent-child relationships, retries, and missing evidence so traces support a careful diagnosis.