Converts noisy log files (up to 100MB) into structured incident reports with a timeline, root cause, and action items — using pre-filtering to cut token usage well before any LLM call happens. The pre-filter-before-you-prompt discipline here is a cost lesson that compounds with the observability project before it: cheaper input means a cheaper, faster pipeline in production.
Incident response tooling sits squarely in platform/SRE-adjacent engineering — proof you can apply AI to operational, not just developer-facing, problems.