Google · Software engineer · Senior · L5+
System design
The L5+ system design round. Drive it: requirements, architecture, tradeoffs, and what breaks at scale.
What they actually ask
The question shapes this round uses. Larpy does not read these out. It builds each one from your own resume, because the generic version is the one you have already rehearsed.
- “Design a URL shortener”
- “Design a rate limiter”
- “Design a news feed or notification system”
- “Design an autocomplete or typeahead system”
- “Design a distributed key-value store or cache”
What this round is judged on
- Requirements · you scope and estimate before you design
- Architecture · a coherent high-level design with a real deep dive
- Tradeoffs · you justify each choice and adapt when constraints change
- Scale · you reason about bottlenecks, capacity and failure
- Communication · you drive the round and stay clear throughout
Where this round sits
Google’s published process. Lime marks the stage this round runs.
- 01Resume review and online assessment
- 02Recruiter screen
- 03Technical phone screen
- 04Onsite loopthis round
- 05Hiring committee
- 06Team match and offer
What happens in each stage›
- Resume review and online assessment
- A human resume screen, for early-career roles often paired with the Google Hiring Assessment, a situational questionnaire rather than a coding test.
- Recruiter screen
- A roughly 30 minute call on your background, the role and level, timeline and comp expectations. Not technical.
- Technical phone screen
- A 45 minute coding round in a shared Google Doc with no autocomplete and no execution. Usually one round, up to two for senior levels.
- Onsite loop
- Five 45 minute rounds. For L3 and L4, four coding rounds plus a Googleyness and leadership round; for L5 and above, three coding rounds, one system design, and Googleyness and leadership. Each round is assigned one or two of the four attributes, and interviewers write up evidence rather than a verdict.
- Hiring committee
- Googlers who never met you read the written packet and make the call. This is why unsupported claims fail here: the room cannot vouch for you, only the notes can.
- Team match and offer
- Matching to a team with headcount, then senior review before an offer is extended.
Sources
Built from what Google publishes about its own process and from real interview data. No leaked question lists. The questions you get are generated against your own resume, so they are not from anyone else's interview.
The four attributes and the hiring committee
SWE loop shape and stage names
How Google tells candidates to structure answers
Why the hiring committee exists and why interviewers write evidence rather than verdicts. An insider account of the structure, though published in 2015, so treat specifics as historical and the structure as current.
Google's own recommended data-structures and algorithms scope for coding interviews
The loop structure and round counts by level, the 45 minute Google-Doc coding format, and the Strong-No-Hire to Strong-Hire scale
Coding-round format and the rubric dimensions (communication, problem-solving, coding, testing), community-maintained and permissively licensed
Sources last checked . Hiring loops change, so this date is the honest limit on everything above.
Other Google rounds
- Behavioral · leadership and Googleyness
The leadership and Googleyness slots of an L3 loop. Judged on evidence a committee who never met you can read.
- Behavioral · leadership and Googleyness
The intern behavioral slot. Heavier on curiosity, humility and how you work with people than on scope.
- Role related knowledge
The RRK slot, spoken not coded. Whether your claimed depth is real when someone digs into it.
- Recruiter screen
The 30 minute recruiter call that opens a Google loop. Background, motivation and level fit, not technical.
- Coding phone screen
The 45 minute coding phone screen, in a bare shared doc. One medium problem, explained before it is written, then follow-ups.
- Onsite coding
An onsite 45 minute coding round. Harder, deeper follow-ups, and the optimality bar is real.