

ram
110 posts

@ramdetwt
Fable of the Dragon Benefactor (mine but it’s not important).







Interviewed a backend dev with 12+ years. They blanked on a few modern things: didn’t know Terraform well, hadn’t used OpenTelemetry, and their answer to async work was still cron + a DB table, not queues. Then we did a debugging exercise. A service was timing out at 2s, CPU was fine, p95 latency spiked, and the DB looked healthy. They asked for logs first, then traced the request path on paper, then found a pattern: only requests hitting one endpoint, only after deploy, only when a feature flag was on. They noticed a missing index in the query plan, but also called out the real root cause: the deploy changed a default pagination size, turning a safe query into a table scan. Their fix included the index, a rollback plan, and a guardrail test. Experience is knowing how to figure things out under pressure, not knowing every library










