At Google, the engineers who interview you do not get to hire you. They write up detailed feedback and a rating, and a separate hiring committee you never meet reviews the packet and makes the call. That shapes the whole loop: candidate reports describe something long, structured and deliberately opaque, where every round is about leaving clear, defensible signal on paper, not winning over one person in the room.
For most software roles it runs recruiter screen, then a technical phone screen (often a single coding problem), then a virtual or onsite loop of two coding rounds plus a behavioural round Google calls Googleyness. Some pipelines add a Google Hiring Assessment early on, and senior, ML, SRE and Forward Deployed roles swap in a system design or domain round. The technical screen has you code in a plain Google Doc that offers no autocomplete and nothing to run or test against, which catches people out almost as much as the problems.
Levels run from L3 (early career) through L4 (SWE III) to L5 and above. Passing the technical bar is not the finish line: you then enter team matching, where a hiring manager has to actually want you, and candidates report sitting in that pool for weeks or months, sometimes with the application closed for lack of headcount.
The timeline is the part almost everyone underestimates, with long gaps between recruiter contact, the phone screen, the loop, the committee and team matching. What follows is each stage in order: the questions candidates said they were actually asked, how to prep for a bar that rewards clean implementation under pressure as much as spotting the pattern, and why the committee and the team-matching queue can matter more than impressing any single interviewer.
Walking into this interview soon? GhostPilot listens to your live call, spots the question the moment it is asked, and puts a structured answer on your screen in real time. Try it on your next mock, or grab a $29 Session Pass, no subscription, for the real thing.
See how it worksThe interview process, round by round
- Recruiter screen Call
A recruiter, often the one who found you on LinkedIn, walks through the pipeline and confirms your target level and location. Candidates in Europe report being nudged to book the first round within about two weeks. Level gets set here, and level decides how hard the loop will be.
- Google Hiring Assessment / OA Online, role-dependent
Not everyone gets one, but many report a Google Hiring Assessment (GHA) or online assessment early on. Some describe it as largely behavioural, others as a coding assessment. New-grad and internship pipelines lean on it most.
- Technical phone screen 45 min, Google Doc
One data-structures problem with follow-ups, coded live in a shared Google Doc: no way to run or test it, no autocomplete to lean on. Reported examples: heap or priority-queue problems, medium graphs, and a two-pointer or greedy problem with an optimisation follow-up.
- Googleyness (behavioural) 45 min
A behavioural round on collaboration, ownership, conflict and decision-making, best answered in STAR format. Many find it conversational, but a No Hire here can sink an otherwise strong packet and trigger a cooldown, and some candidates get a Googleyness retake.
- The onsite loop: coding rounds 2 x 45 min, back to back
Typically two coding rounds, in person at Mountain View or New York, or fully virtual. Forty-five minutes each, one medium-to-hard problem apiece, typed into a Google Doc. Google writes many of its own problems, so pattern-matching only carries you so far.
- System design or domain round 45 to 60 min
For L5 and up, and for ML, SRE, infrastructure, data and Forward Deployed roles. Reported flavours include ML system design (for example designing an AI coding system), infrastructure design, a Forward Deployed 'Role Related Knowledge' round on agentic AI, and, in one report, a standalone technical SQL round.
- Hiring committee and team matching After the loop
Interviewers submit ratings (Strong Hire, Hire, Lean Hire, Lean No Hire, No Hire) into a packet, and a hiring committee you never meet makes the call, sometimes requesting an extra round. Pass, and you move to team matching to find a team with headcount, which can take weeks or months.
Interview questions candidates report
Coding and data structures
- Count the paths through a grid with restricted moves (dynamic programming)
- Grid DFS and backtracking: the longest valid path under movement constraints
- Trie or prefix search: given words and a prefix, return every word that starts with it
- Rounds-based array ops: scan left or right for the nearest value that is double the current one
- Heap and priority-queue scheduling, plus 'minimum meeting rooms' interval problems
- Weighted graph (a path from u to v minimising the maximum edge weight), tree, string-matching (KMP) and combinatorics problems
Follow-ups and optimisation
- Space-optimise a working DP solution (often not expected without a hint)
- Add extra constraints to a solved problem and re-derive the approach
- State time and space complexity precisely, including amortised cost
System design and domain rounds
- ML system design, for example designing an AI coding system
- Infrastructure and general system design for L5 and above
- Forward Deployed 'Role Related Knowledge': design with an agentic-AI focus
- Python OOP and object-oriented design for FDE-style coding rounds, plus a standalone SQL round in one report
Googleyness and behavioural
- Walk through a time you navigated conflict while collaborating with a teammate
- Describe a decision you made and how you weighed the trade-offs
- A time you took ownership of something that was not strictly yours
Product Manager variant
- Product design: 'Design a product for elderly users' (clarify goals and segments first)
- Product sense, technical, leadership and Googleyness across four back-to-back rounds
What candidates said about the process
A product manager said the recruiter laid out the format up front: a phone screen, then a four-round onsite of 45 minutes per round spanning product sense, a technical round, leadership and Googleyness, and finally the hiring committee. The phone-screen prompt was to design a product for elderly users, and they credited passing to clarifying goals and segments before proposing anything, after roughly six weeks of prep and thirty mock interviews.
Read the candidate's account →An early-career candidate in Europe shared a telling timeline: recruiter contact, a call a week later, a medium graph phone screen a month in, then a behavioural round that only landed on the fourth attempt, after two interviewer no-shows and a technical round that began before being cancelled partway. After positive feedback they sat two DSA rounds onsite and moved to team matching, over four months in.
Read the candidate's account →One candidate cleared both phone screens, a rounds-based array problem (scan left or right for the nearest value that is double the current one) and a conversational Googleyness round, but was rejected after the onsite. Their diagnosis was blunt: recognition was never the gap, implementation speed under pressure was. They spotted an onsite Trie prefix-search problem instantly and still could not code it cleanly in time.
Read the candidate's account →An ML-focused L4 candidate was offered a choice at the recruiter stage between ML system design and DSA, and picked ML. After a Google Hiring Assessment, the first loop paired a Googleyness round (more discussion than interrogation) with a design round that turned out to be designing an AI coding system, not the domain quiz they expected. The recruiter graded each round only Positive, Mixed or Negative; both came back Positive.
Read the candidate's account →A cautionary one: a candidate cleared Googleyness, an AI/ML round plus a pair of coding rounds, matched with a team, and had started comp talks when the hiring committee flagged that one interviewer had spent the slot on resume chat, leaving too little signal. Two extra rounds followed, then a rejection for unspecified 'confidential reasons' with a six-month cooldown. The committee, not the room, decides.
Read the candidate's account →How to prepare
- Train implementation speed, not just recognition. The common failure report is not missing the pattern but running out of clock coding it, so practise typing full, correct solutions under time pressure.
- Rehearse in a bare Google Doc. Coding with no autocomplete and no run button is its own skill; make sure the format is not the thing that throws you on the day.
- Say your complexity out loud and get it right. Candidates were marked down for shaky time and space analysis even when the code worked, so be ready to reason about amortised cost too.
- Do not treat Googleyness as a warm-up. A single No Hire there can close your application with a cooldown. Prepare concrete STAR stories on collaboration, conflict and decision-making, one per question.
- Expect follow-ups and hints. Interviewers push a working solution toward better space or extra constraints; needing a hint is not fatal, but stalling in silence is, so keep narrating.
- Plan for a long, stop-start process. Passing the loop only gets you into team matching, so keep your problems warm through the gaps and do not bank on a fast decision.
Frequently asked questions
How many rounds is the Google interview?
Expect a recruiter screen, a technical phone screen, and an onsite loop of two coding rounds plus a Googleyness round, with a system design or domain round for senior roles. A hiring committee then reviews the written feedback and makes the final call.
What is the Googleyness round?
A behavioural round on collaboration, ownership, conflict and decision-making, best answered in STAR format. Candidates often find it conversational, but it carries real weight: a No Hire here can sink a strong technical packet and trigger a cooldown.
Do you really code in a Google Doc?
Yes, at least for the coding screen. One person described writing their solution in a plain Google Doc: no autocomplete, and nothing to run or test as you go. Another, heading into an onsite, asked whether that stage uses a Doc or a whiteboard, so plan to practise away from a full IDE either way.
What is team matching and the hiring committee?
Interviewers do not hire you; they submit ratings into a packet a hiring committee reviews. Pass, and you enter team matching to find a team with open headcount, which can take weeks or months and occasionally ends without an offer despite a passing loop.
What coding questions does Google ask?
Reported examples include grid dynamic programming, DFS and backtracking paths, Trie prefix search, heap and priority-queue scheduling, weighted-graph variants, trees, and string matching. Google writes many of its own, so expect unfamiliar framing and hint-driven follow-ups.
How long does the Google interview process take?
Usually a long time, stretching across weeks or months with gaps between the recruiter call, phone screen, loop, committee and team matching, plus the odd reschedule. Interview scores tend to stay valid for roughly six to eighteen months if things stall.
Google will ask its own version of these. Paste your actual job description into the free Question Predictor and get the 20 questions that role is most likely to ask, with what each one is really probing.
Predict my questions