The short answer: Site reliability engineers keep large services available, fast and cheap to run, and they do it by writing software. The interviews mix coding with rounds most software engineer loops do not have: live troubleshooting, Linux and networking internals, and reliability design with real numbers. Only Google calls the role Site Reliability Engineer across its postings. Meta hires Production Engineers, and Amazon hires mostly Systems Development Engineers. None of the three publishes a current SRE-specific loop. Google's most detailed description is a 2015 article by its SREs, and the round mix at all three comes from candidate reports, so ask your recruiter for your exact rounds. Last reviewed October 2026.
Site Reliability Engineer interview: quick answers
| Question | Short answer |
|---|---|
| What does a site reliability engineer do? | Runs production services and writes the software that keeps them reliable: automation, monitoring, capacity plans, incident response and postmortems. |
| What do the interviews test? | Coding, troubleshooting a live-style incident, Linux and networking internals, reliability design with numbers, and behavioral stories. Google calls its design round NALSD. |
| Can I use AI tools? | Google says no. Meta allows no outside AI tools, only the tools inside the interview environment. Amazon publishes no rule, so ask your recruiter. |
| How much do SREs make? | Posted US base ranges run from $99.1K for Amazon SysDE I to $364K for Google Senior Staff SRE. Meta's two Production Engineer postings list $122K to $257K. |
| How is it different from a software engineer interview? | The coding bar is similar, and candidate reports say some SRE loops look almost the same. The difference is the systems rounds: troubleshooting, internals and design measured in machines and bytes. |
| Is there an entry-level path? | Meta posts Production Engineer internships and Amazon posts SysDE I and intern roles. Google had no US intern or early-career SRE posting on October 28, 2026. |
What site reliability engineers do
Google's postings describe SRE as what you get when you treat operations as a software problem. SRE teams have full authority to fix and scale the code they run, and Google caps their operational work at 50 percent so the rest goes to engineering. The work spans launch reviews, monitoring, automation, incident response and capacity planning for products such as Search, Ads, YouTube and Google Cloud.
Meta's Production Engineers are specialized software engineers who own back-end services, write capacity plans and share an on-call rotation with their team. They sit in product teams such as Instagram and WhatsApp and in core services such as Storage, Cache and Networking.
At Amazon the same work lives mostly in the Systems Development Engineer job family, much of it on AWS teams such as EC2, DynamoDB and Lambda. The one US posting titled Site Reliability Engineer, in Amazon Ads, owns SLOs, error budgets and operational readiness reviews, and drives the Correction of Error process after incidents.
How 3 companies run the site reliability engineer interview
The short version of each process. Each company guide has the full detail, with links to the company's own pages, and says which parts come from candidate reports.
| Company | Screens | Main interviews | AI tools | What stands out |
|---|---|---|---|---|
| Recruiter calls, then a technical phone screen | 4 or 5 interviews: coding, NALSD, troubleshooting, internals (2015 article) | Not permitted | Hiring committee decides; team match comes after | |
| Meta | Coding and troubleshooting screens (candidate reports) | Coding, design, Linux internals, behavioral (candidate reports) | Only tools inside the interview environment | The role is titled Production Engineer |
| Amazon | Phone screen; assessment for some roles | Loop of 2 to 7 interviewers; Leadership Principles in every round | No published rule | The role is mostly titled Systems Development Engineer |
The interview types in depth
Troubleshooting
You get a vague symptom, such as a website being down or errors after a deployment, and work it with the interviewer as details change. Candidate reports describe this round at Google and in Meta's screen. You are judged on how you narrow the problem down step by step.
A structure that works: confirm the impact and ask what changed, mitigate first, then list hypotheses and test each with a named command and an expected output. Finish with the root cause, the short-term fix and the long-term fix that stops it happening again.
Reliability design and NALSD
SRE design rounds want numbers. Google's NALSD (non-abstract large system design) asks for realistic estimates of throughput, storage and machines for each component, and its SREs wrote in 2015 that naming a magic technology is not enough. Meta's Production Engineers write capacity plans, and candidate reports describe a PE design round. Amazon's technical topics include distributed computing and system design.
Start with requirements and an SLO, then size the system: requests per second, bytes per request, storage per day, then machines with headroom. Name the failure domains (host, rack, zone, region) and show what happens when each fails. Close with how you would deploy changes safely and what you would alarm on.
Linux and networking internals
This is a technical conversation that goes deeper until you run out of answers. Candidate reports describe processes and signals, memory management, file systems and links, and what happens on the network when you type a URL. Google's 2015 article lists UNIX internals and networks as interview topics, and a 2025 Amazon SysDE candidate reported a Linux and networking round.
Hands-on practice helps most here. Break a test machine on purpose (fill a disk, leak file descriptors, drop packets) and find the cause with standard tools, so you can explain each command and its output.
Coding
SRE coding rounds lean practical. Candidate reports describe log processing and file parsing at Google and Meta, with follow-ups on bad input, files larger than memory and tests. Some Google software engineering track candidates report classic algorithm problems instead, and a 2025 Amazon candidate said coding felt closer to low-level design.
Write a correct version first, then improve it as the interviewer adds requirements. Say how you would test it and what you would log, since that is how operational code fails in practice.
Behavioral
Every loop has behavioral questions, and SRE versions lean on incidents, postmortems and work you automated away. Amazon evaluates every candidate against its Leadership Principles in every interview. Google suggests STAR answers with measured impact.
Bring a small set of strong stories: an outage you handled, a postmortem or Correction of Error you wrote, toil you removed, and a disagreement about a launch or a reliability risk.
Pay in US job postings
US salary ranges from each company's own postings for this role, seen October 29, 2026. They are base salary unless the company guide says the posting lists total salary. Each company guide has the postings and links.
| Company | Titles in the postings | Posted US pay range |
|---|---|---|
| Software Engineer III, Senior Software Engineer and Staff Software Engineer, Site Reliability Engineering; Staff and Senior Staff Site Reliability Engineer; Site Reliability Engineer, SWE, Cloud | $147K to $364K | |
| Meta | Production Engineer and Production Engineering (Meta posts no Site Reliability Engineer title) | $122K to $257K |
| Amazon | Systems Development Engineer (SysDE I, II, III and Senior), also spelled System Development Engineer; one US posting titled Site Reliability Engineer, Amazon Ads PubTech | $99.1K to $235.2K |
Entry-level paths
- Meta: Meta posts a Production Engineer Intern role for Bellevue, Menlo Park and New York and another for London; Network Production Engineer (University Grad) is posted in Dublin and London.
- Amazon: Amazon posts SysDE I roles, a System Development Engineer Intern/Co-op role in Robotics for 2027, and a university interview page that covers SysDE.
- Google: Google had no US intern or early-career SRE posting on October 28, 2026; the lowest US tier asks for 2 years of software development.
AI tools in site reliability engineer interviews
Rules on AI differ by company, so ask your recruiter before every round.
Google. Google says AI tools are not permitted during interviews, and its tips page says the same of Google Search and AI.
Meta. Meta says no outside AI tools or assistance are authorized, and only the tools available within the interview environment may be used. Its FAQ says many interviews for select roles include a built-in AI assistant that candidates are expected to use, framed around software engineering skills. It does not say whether Production Engineer interviews use it.
Amazon. Amazon does not publish a rule on AI use in interviews. A March 2025 news report described internal Amazon guidance asking candidates not to use generative AI tools during interviews, which is a report and not a published Amazon rule.
A 6-week prep plan
| When | Focus |
|---|---|
| Week 1 | Read your target company's postings and guide. Note its title for the role, its teams, its AI rule and which rounds candidate reports describe. Start reading the SRE books on sre.google. |
| Week 2 | Linux and networking: processes and signals, memory, file systems, TCP, DNS, HTTP and TLS. Break a test machine and find each fault with standard tools. |
| Week 3 | Troubleshooting: one incident scenario a day, worked out loud from symptom to root cause and fixes. |
| Week 4 | Reliability design: two problems a week with full numbers, failure domains, SLOs and an error budget. |
| Week 5 | Practical coding: log and file parsing, small automation scripts and classic data structure problems, with no AI help. |
| Week 6 | Behavioral stories and mock loops: an outage, a postmortem, toil removed and a disagreement, then two full practice loops. |
Courses for each round
Practice questions
Questions from our bank, then practice prompts from the company guides.
Questions from our bank
From the DataInterview question bank, picked for site reliability engineer interviews. Open one to answer it.
Latency Issue Mitigation
Internet Operation
Kubernetes Control vs Data Plane
File Downloader Design
System Architecture Experience
Counting Likes for Influencers
API Rate Limiter Design
Cross Region Inventory Sync
Caching Overview
Troubleshooting practice
One prompt per company, from the company guides. These are our prompts, not reported interview questions.
A Search frontend cluster's error rate rose from 0.1 to 2 percent, with timeouts, ten minutes after a new binary started rolling out. Walk through what you do.
Sample Answer
Mitigate first. The SRE book says about 70 percent of outages come from changes, so pause the rollout and roll back the affected cluster while you investigate. Confirm the impact on users through the error and latency dashboards. Then compare old and new tasks: error logs, CPU and memory, and calls to backends. Check whether the timeouts are on one backend dependency or across the board. A strong answer ends with the root cause, the short-term fix (rollback) and a long-term fix such as a canary stage that would have caught it.
Users say a Meta web property is down. You have shell access to the web tier. Walk through how you find the cause.
SSH to an EC2 host works, but the load balancer marks the service on it unhealthy. Walk through what you check.
Reliability design practice
One prompt per company, from the company guides. These are our prompts, not reported interview questions.
NALSD: join query logs and click logs written in three datacenters. Each datacenter logs 100,000 queries per second at 1 KB each, 1 percent of queries get a click, and a click arrives within an hour of its query. Size the join.
Sample Answer
Total query traffic is 300,000 per second, or 300 MB/s and about 25.9 TB a day (300 MB/s x 86,400 s). Clicks run at 3,000 per second. The join must hold one hour of queries keyed by query id: 300 MB/s x 3,600 s = 1.08 TB, or 3.24 TB with three replicas. At three replicas that is 900,000 writes per second; if one storage server handles 50,000 writes per second, you need 18 servers before headroom, holding about 180 GB each, so throughput sets the server count. Storing only the fields the join needs shrinks storage further. Cover what happens when one datacenter is unreachable for an hour.
Plan capacity for a photo upload service: 20,000 uploads per second at peak, 500 KB average, three stored copies, and average traffic at half of peak.
A regional service handles 50,000 requests per second with a 99.9 percent monthly availability target. Design its alarms.
Fundamentals practice
One prompt per company, from the company guides. These are our prompts, not reported interview questions.
What is the difference between a hard link and a soft link, and what happens to each when you delete the original file?
Sample Answer
A hard link is another directory entry for the same inode, so the data stays until the last link is removed. A soft (symbolic) link is a separate file that stores a path, so deleting the original leaves it dangling. Hard links cannot cross file systems or usually point to directories; soft links can. A strong answer adds that a deleted file held open by a process still uses disk space until the process closes it.
What happens when a Linux machine runs out of memory?
Walk through the TCP three-way handshake. What does it look like when a server's accept queue overflows?
Behavioral practice
Prompts we wrote for site reliability engineer behavioral rounds, with what a strong answer covers.
Tell me about an outage you handled from the first alert to the postmortem.
Sample Answer
Describe the impact in numbers, how you mitigated before you knew the cause, how you found the root cause and the action items you owned. Keep it blameless and say what you would do faster next time.
Describe toil you removed through automation.
Tell me about a time you pushed back on a launch for reliability reasons.
Tell me about a time you debugged a system you did not know.
Are you ready for site reliability engineer interviews?
1 / 5Can you debug a slow or failing Linux service live, from symptoms to root cause, while explaining each command?
Company guides
Read the guide for each company you are interviewing with:
- Google: The Google SRE interview: the official hiring steps, the 2015 loop Google SREs described, what candidate reports add, the AI rule and pay in Google's postings.
- Meta: Meta hires SREs as Production Engineers. The official hiring steps, what candidate reports say about each round, the AI rule and pay in Meta's postings.
- Amazon: Amazon hires most SREs as Systems Development Engineers. The official hiring steps, what candidate reports add, the AI question and pay in Amazon's postings.
Frequently Asked Questions
What is a site reliability engineer interview like?
Expect coding plus systems rounds: troubleshooting a simulated incident, Linux and networking internals, reliability design with numbers, and behavioral questions. The exact mix depends on the company and team, and most of what is known about current loops comes from candidate reports.
Does every company call the role Site Reliability Engineer?
No. Google uses Site Reliability Engineering titles. Meta's SRE-type role is the Production Engineer, and Amazon's is mostly the Systems Development Engineer, with one US posting titled Site Reliability Engineer.
What is NALSD?
Non-abstract large system design, Google's SRE design interview. You design a concrete system and size each component with realistic throughput, storage and machine counts. The SRE workbook on sre.google has a chapter on it.
Can I use AI tools in SRE interviews?
Google does not permit AI tools. Meta allows no outside AI tools, only the tools inside the interview environment. Amazon publishes no rule, so ask your recruiter.
How much do site reliability engineers make at Google, Meta and Amazon?
In US postings seen in October 2026, Google lists $147K to $364K base from Software Engineer III to Senior Staff, Meta lists $122K to $257K for Production Engineers, and Amazon lists $99.1K to $235.2K for SysDE I to Senior SysDE. All three add bonus or equity.
How long does the process take?
Google says about 6 to 8 weeks once interviews start. Meta says about 2 to 3 months. Amazon says to expect a decision within five business days of the loop.
Is SRE harder to get into than software engineering?
The coding bar is similar. Google's SREs wrote in 2015 that its software engineering bar inside SRE matches product development. The extra work is the systems depth, which takes hands-on Linux and networking practice to build.
