Site reliability engineer phone screen questions: SLOs, incidents, on-call load, toil and code

A site reliability engineer phone screen is a 20 to 30 minute recruiter call that checks whether the candidate has owned the reliability of real services, not only kept them running. Confirm which services they were responsible for and whether those had service level objectives (SLOs) they helped set, the on-call rotation and how many pages it produced, the language they write code in every week, and logistics. You do not need to judge the engineering. Listen for numbers only the owner would know: the SLO, the last time the error budget ran out, the incident they led and the manual work they automated away.

This page is the first-call version. The deeper bank, with error budget arithmetic, incident command and database reliability seats, is in site reliability engineer screening questions. If the seat is mostly pipelines and infrastructure code, the DevOps engineer phone screen fits better; the ownership ladder for hearing coding depth is in software engineer phone screen questions, and the general call structure is in phone screen questions for recruiters.

What do you need from the hiring manager before the call?

"SRE" can mean an engineer embedded with one product, a platform team, an observability team, database reliability or an operations team with a new name. Get these facts in the intake:

How should a 25-minute SRE phone screen run?

1. Opening (1 to 2 minutes). Who you are, the seat in two sentences, the length of the call, and your consent line if you record.
2. Knockouts (4 to 5 minutes). On-call, SLO ownership, weekly coding, primary platform, location, work authorization.
3. Reliability snapshot (7 minutes). One service: its objective, the budget, the last incident, the toil.
4. On-call and toil in numbers (3 minutes). Pages, hours and the share of the week.
5. Motivation (3 minutes). Why now and what kind of reliability work they want next.
6. Pay expectations and logistics (3 minutes). Base, equity, on-call terms, notice, other processes.
7. Their questions and the next step (2 minutes). Consent to submit, availability for the technical rounds.

The timings are a planning example. If a knockout fails, say so kindly and end the call or redirect.

Which knockout questions come first?

The authorization wording follows the Justice Department's Immigrant and Employee Rights Section FAQ, which says an employer may generally ask whether applicants have the legal right to work in the United States and whether they will need visa sponsorship (as of November 2026).

What is the reliability snapshot?

One service, four questions, about seven minutes. It replaces a tour of the candidate's tool list with a story the candidate cannot look up.

1. "Pick one service you were responsible for. Who used it, and what was its SLO?" Good: users, a target and how it was measured. Weak: "It was always up."
2. "When did that service last run out of error budget, and what changed because of it?" Good: a date, the cause, and a consequence such as a release freeze or reliability work moved ahead of features. Weak: the budget existed on a dashboard and nothing ever changed.
3. "What was the last incident you led or were paged for? What was your role, and what did the postmortem change?" Good: their role (incident commander, communications, hands on keyboard), a short timeline and one action item that was actually done. Weak: postmortems written, no follow-up ever finished.
4. "Which manual task did you automate away, and how much time did it give back?" Good: the task, the tool they wrote and a rough number of hours a week. Weak: "We automate everything."

The example answers are invented; the pattern is what matters. To follow them you need three terms from Google's SRE book, as of November 2026. An SLO is a target value for a service level measured by an indicator (an SLI). Google's example error budget policy defines the budget as 1 minus the SLO and halts changes other than P0 issues or security fixes once it is spent. Example arithmetic: a 99.5 percent availability SLO over 30 days allows 0.5 percent of 43,200 minutes, or 216 minutes. And Google says blameless postmortems are a tenet of SRE culture, so a candidate who tells an incident as someone else's fault is worth one more question.

How do you hear on-call load and toil in numbers?

Ask for numbers, not adjectives. Google's SRE book gives reference points, as of November 2026: SREs should spend at least 50 percent of their time on engineering and no more than 25 percent on-call, with at most two incidents per 12-hour shift because one incident and its follow-up take about six hours, and Google aims to keep toil (manual, repetitive, automatable work that grows with the service) below half of each SRE's time. These are Google's own targets, not a rule for your client; use them to read the answers.

Which certifications should you check?

SRE seats rarely require a certification unless a contract or partner program does. When one is required, ask for the issuer's verification link or badge and the expiry date. The details below come from each issuer's page, as of November 2026; the AWS and Microsoft equivalents are in the DevOps engineer phone screen.

A current certification plus a reliability snapshot full of numbers is the strongest combination. Certifications without an owned service are common among career changers and are a reason to ask more questions, not to reject.

How do you test motivation on an SRE phone screen?

How do you ask about pay without asking salary history?

Many states and cities bar employers and their agents from asking about a candidate's current or past pay, as of November 2026; the list is in salary history ban states. Ask what the candidate needs from the new role.

If the candidate volunteers a number, keep it out of your notes and the submittal and record the expectation instead. This is general information, not legal advice; check the law where the job is located.

When should you advance a site reliability engineer?

An example rule to agree with the hiring manager: advance when every knockout passes, the reliability snapshot scores at least 2, and two other areas score 3. Hold when one fact, such as a certification's expiry, needs checking. Decline when a knockout fails. The engineers' rounds are in the interview guide for DevOps roles, and once hired, the 30-60-90 day plan for site reliability engineers starts with the service inventory, the SLO baseline and on-call shadowing.

What should you not ask an SRE on a phone screen?

The EEOC says pre-employment inquiries should be limited to what is needed to judge whether someone is qualified, that questions which reveal race, sex, national origin, age or religion may be used as evidence of intent to discriminate, and that employers may not ask about disability before an offer. On-call is where SRE screens drift into unlawful territory; ask about the schedule, not the candidate's home life.

If you take the screen through your computer, Interview Signal keeps the on-call, SLO and incident questions pinned until you ask them and writes the candidate's own words beside each score, with no bot in the call.