Incident and Escalation Manager

Vapi•San Francisco, CA

13h•$180,000 - $250,000

About The Position

Vapi runs voice AI infrastructure at tens of millions of minutes a month for enterprise customers who route business-critical traffic through us. When something breaks, it breaks in real time, in front of a customer's own customers, and often through a carrier dependency we don't fully control. Today the response is run by whoever is in the Slack thread. That works once. It does not scale, and it costs us renewals, engineering focus, and executive confidence every time it happens. This is the first dedicated hire for the function, and the job is to build it. You will design the program that runs our incident response, train the people who command incidents, and own the customer relationship through the weeks that follow a serious one. You own the pager and you work part of the rotation yourself, because the fastest way to build a program that works is to run inside it while you build it. You are not the engineer fixing the system. You are the person who builds the system so a trained bench can run it, and who takes a real turn on the pager.

Requirements

8 to 12 years across incident management, escalation management, technical support escalations, or technical program management, ideally at an infrastructure, telephony, or platform company operating at scale.
A track record building an incident and escalation program from zero, or owning a meaningful piece of one through its growth, including standing up an incident commander rotation rather than being the sole responder. Experience hiring a team is a plus.
Hands-on familiarity with incident tooling such as PagerDuty, incident.io, Opsgenie, or equivalent, and the Slack and status-page workflows around them.
Calm under pressure as a learned discipline. Gravitas to direct a response and the willingness to remove a distraction from a call even when it outranks you.
Decision-making with incomplete information, and the judgment to know when to escalate and how to do it without losing time.
Clear writing under pressure. You can produce a clean read of a live situation for a senior leader, and a customer RCA that holds a relationship together.
Comfort with technical depth. You do not need to write the fix, but you need to follow the conversation, ask the right question, and know when an answer does not add up. Familiarity with telephony, carrier dynamics, or real-time systems is a strong plus.
Willingness to work part of the incident commander rotation, including off-hours shifts, especially in the first year.

Nice To Haves

Experience hiring a team is a plus.

Responsibilities

Define what counts as an incident and what does not, with entrance and exit criteria that protect engineering from noise.
Author the severity model with response-time targets per level, so a SEV0 means the same thing Monday morning and Friday night.
Design how a live incident runs: the command structure, the update cadence, the single source of truth, the decision rights, and when the call gets escalated past the commander. Write it down so anyone on rotation runs it the same way.
Build and train the incident commander rotation. Build a realistic balance of hiring and utilizing existing resources across humans and AI to stand up the rotation.
Own the pager and work part of the rotation yourself - personally commanding incidents during your shifts and stepping your share down as the bench matures and proves out. Own how the rotation runs regardless of whose shift it is.
Stand up the incident tooling and on-call setup: paging, escalation policies, incident channels, status page, and the runbook library.
Build customer communication templates for each severity and channel, pre-approved so they are not written from scratch under pressure.
Govern the customer credit process with a clear approval chain, so financial decisions stop happening in ad hoc threads.
Stand up the metrics: resolution time, response velocity, escalation volume, RCA SLA adherence, and revenue protected. Report monthly in terms execs use to make resource decisions.
Build standing partnerships with engineering, support, the office of the CTO, legal, security, comms, and carrier operations before the next critical situation.
Train go-to-market, support, customer success, and engineering on where to bring customer-critical issues and how the function works.
Build a feedback loop so incident and escalation data shapes the engineering roadmap instead of dying in postmortems.
Own the customer-facing RCA. Translate engineering root cause into plain language that tells the truth and holds the relationship. Ship it within the SLA we commit to.
Run the blameless post-incident review. Drive action items to named owners with dates, and track them to closure instead of letting them die in a doc.
Close the loop with affected customers directly, including the credits or commitments made during the incident.
Hold the high-severity customer issues that do not rise to a full incident but threaten a renewal or a relationship.
Run the standing executive escalation list. Keep an owner, a next step, and a date on every item.
Spot patterns across customers that no single team owns, and force them into engineering or product as prioritized work.
Be the single point of contact for a regulated customer or a regulatory inquiry that surfaces weeks after the technical fix.
Partner with the account team on at-risk accounts driven by reliability, and build the cross-functional recovery plan.
Keep a written handoff and a named deputy so escalations stay covered when you are out.