Her production pipeline fails at 2 a.m. in the morning. You are the on‑call engineer. What is the first thing you will do?

The first thing I do is confirm the incident and gather context. I check the alerting system to verify the failure and note the exact error message or status code. Next, I look at the pipeline logs for the most recent run, focusing on the point of failure. I also review any recent code or configuration changes that could have triggered the issue. If the pipeline is orchestrated by a scheduler (e.g., Airflow, Prefect), I check the scheduler’s health and any dependency services (databases, storage, APIs). I then attempt to replay the failing step in a safe environment or run a dry‑run to isolate the problem. While doing this, I communicate with the team via the incident channel, providing status updates and asking for any recent deployments or incidents that might be related. If the failure is due to a resource constraint (CPU, memory, network), I scale the resources or restart the affected workers. Finally, I document the root cause, fix the issue, and create a post‑mortem to prevent recurrence.