Case 001 · Priority One
A playable SRE detective experience. Follow the telemetry, separate coincidence from cause, and restore checkout before the trail goes cold.
Open the case file →Live simulation
Opening briefing · 22:14 UTC
At 22:04, traffic surged. Since then, checkout success has fallen from 99.6% to 57.8%, and P95 latency climbed from 420 ms to 8.4 seconds. Inventory appears healthy. Database connection errors are present in checkout logs.
Behind the build
This prototype turns incident response into a guided customer-evaluation asset. The browser runs deterministic fixtures; a production implementation would replace them with live, permissioned telemetry.
Live implementation
Prototype boundaries
Recruiter-friendly skills map
Extend the system
Create a typed scenario fixture with timeline, evidence, suspects, remediation options, and scoring keys. The investigation shell remains reusable.
app/scenario.ts → ScenarioSolutions Engineer mode
DiscoverHow do you prioritize signal quality and reduce mean time to recovery?
DemonstrateShow causal correlation, SLO impact, and a repeatable investigation path.
DiscoverWhere do teams lose context across metrics, logs, traces, and cluster state?
DemonstrateShow OpenTelemetry portability, service topology, and Kubernetes evidence.
DiscoverWhat is the business cost of slow, inconsistent incident response?
DemonstrateConnect checkout recovery to revenue protection and operational learning.
Suggested path
Pilot success criteria
Objections & responses
Dashboards show data. This motion tests whether teams can connect impact, cause, action, and recovery across signals.
Start with deterministic fixtures, then use scoped read-only access and sanitized pilot services.
The assistant is evidence-grounded, cannot reveal unseen data, and leaves remediation approval with the operator.
Run a 45-minute architecture workshop → select one business-critical service → agree on pilot signals and success criteria → schedule a scored incident exercise.
Run the case again →