Incident Operations Commander

Serve as the on-duty commander for Alpaca's most critical incidents, directing cross-functional response to restore service quickly, keeping the right people engaged and informed, and making sure every incident leaves behind something the organisation can act on.

You do not fix the outage. You make the response reliable: correct severity, the right engineers in the room, mitigation that does not stall, leaders informed in time, and follow-up work that survives the call.

Things You Get To Do

Command incidents end to end. Take command from declaration to mitigation, keeping responders focused on stopping customer and partner impact as fast as possible. Run the bridge, keep observers out of the responders' way, and name a stall out loud when you see one.

Classify and hold the line on severity. Set severity at declaration and re-check it as facts arrive. Risk advises on financial and regulatory materiality; the call is yours.

Engage the right people, fast. Identify the owning team by service, symptom and blast radius, page them, and expand the responder set the moment the first team is wrong or not enough. When a page goes unanswered, escalate - and escalate the escalation. Bring in the leaders who must make business calls: feature flags, traffic shedding, failover, freeze-or-ship.

Hold the bridge, and protect the people fixing it. Keep engineering and technical support uninterrupted - questions from stakeholders, partners and executives come to you. Be the single source of truth to the partner communications team on impact, severity and timing: you decide when a status page update or partner contact is needed, they write and send it, and chasing a late or stale update is yours.

Run follow-the-sun handoffs. Deliver warm, high-fidelity handoffs across regions: current impact and severity, mitigation path and next actions, who is in the room, outstanding decisions, and what must not be dropped. The incoming commander confirms ownership before you step away - command never goes dark at a region boundary.

Close the loop, on the clock. Maintain the timeline of facts as the incident runs rather than reconstructing it afterwards - in a regulated business that record has to hold up long after the call ends. Once mitigated, make sure a blameless retrospective is scheduled with a named owner and a timebox, and record where the cause sits - that choice sets which follow-up items are mandatory. Every action item needs a real ticket, one named accountable, a priority and a category, delivered inside the agreed service level. If a postmortem produces nothing but low-priority items, treat that as a signal the analysis stopped early and escalate to SRE rather than passing it on.