AWS Observability Analytics
Day job. I work on services that ingest, store, and query telemetry at AWS scale, and on the console surfaces engineers use to investigate incidents. Specifics stay internal.
The public demo is intentionally synthetic: its service names, metric values, trace spans, deploy event, and failure mode are invented. It shows the shape of the work without describing internal architecture or exposing operational data.
Scrub through the incident and the same regression appears in three connected views. Aggregate latency rises while traffic remains flat; the service map localizes the change; representative traces reveal repeated calls; and a correlated configuration event gives the investigation a concrete hypothesis to test. This is the workflow I care about—turning telemetry from a pile of charts into evidence an engineer can act on.