As of October 13, 2026, these are the 15 additional U.S. Google intern interview reports we’ve found and checked for this update.

Bachelor’s, master’s, and PhD candidates share the SWE/SDE intern question bank. Degree labels below are background information; the topic counts cover all three groups together.

What came up

Trees and graphs were the largest group in this batch. Thirteen of the 15 reports named at least one coding topic.

DFS or BFS was explicitly named in five reports. Topics mentioned once each included Top-K with heaps and buckets, preprocessing and binary search, combinatorics, a Latin square variation, divide and conquer, in-place cyclic sorting, and a Task Scheduler variation.

Counts are by candidate, and one report can cover several topics.

The follow-up pattern: Getting a working solution was often followed by a request to improve time or space complexity, handle another constraint, or walk through edge cases. Several candidates found bugs or a different interpretation of the question during a dry run.

Who reached team matching?

Using Team Match / Project Search as the technical-interview pass marker, 7 of 15 candidates (46.7%) reported passing. The other eight had not shared a confirmed result, so they are marked unknown.

The reports below keep each candidate’s dates and role as shared. Report dates and interview dates are listed separately when both are available.

Candidate reports

23 An iterator then a lake counting problem

Report date: Oct 13, 2026 · Role: SWE PhD Intern

Round 1

The interviewer went straight into coding, with no behavioral questions. The candidate found the interviewer relaxed and the task simple, but the setup confusing. They spent a long time checking what they were being asked to do.

They had to write an iterator using an Iterate class. Only leaf nodes had values. Given a sequence of paths, each next() call had to output one character from the value at a path’s final leaf.

Round 2

This interviewer felt more serious and strict. The round began with a five-minute introduction, followed by coding and time for questions.

The task was to count lakes inside islands. The candidate tried two DFS passes: find water regions through land, separate lake water from sea water, then subtract one from the total.

Afterwards, they felt they’d got confused during a simple problem. They thought the counting condition should have used visited, rather than the DFS return value.

The author titled the post as an unsuccessful interview experience. They did not share a final recruiting decision.

24 Twenty minutes reading a tree merging question

Report date: Oct 13, 2026 · Role: SWE PhD Intern

Round 1

The interviewer was friendly. The tree-merging question had five detailed rules. Understanding the prompt and examples took about 20 minutes.

A screenshot the candidate confirmed showed nodes with names, integer values, and ordered children. The merged tree followed the first tree’s order and used the second tree’s values, matching children by name. The screenshot cuts off during the rules.

The candidate immediately thought of DFS. After another 20 minutes, they’d explained their approach and written the base case and two loops, but hadn’t finished the matching logic. They kept rereading examples while coding.

They hadn’t practiced this question before. In the comments, they suggested DFS plus hashing for duplicate nodes. They described the code as simpler than LeetCode 1932, despite a much longer prompt.

Round 2

They said they solved a greedy question quickly but ran out of time for its follow-up. Neither prompt was shared in the saved comments.

The author titled the post as an unsuccessful interview experience. They did not share a final recruiting decision.

25 Two medium questions then team matching

Report date: Oct 13, 2026 · Interviews: Oct 5 · Role: Summer 2027 SWE Intern · Status: Team matching

This candidate, a master’s student, shared the following timeline:

• Aug 31: Applied.

• Sep 8: Online assessment.

• Sep 16: Interview scheduling.

• Oct 5: Two back-to-back virtual interviews.

• Oct 9: Entered team matching.

They described both coding questions as medium difficulty: one involved a tree, and the other used hashing and a queue. They also answered questions about their projects and how they use AI. The exact coding prompts weren’t shared.

Their previous internship involved a video platform, Spring Boot microservices, RocketMQ, Redis, Elasticsearch and MySQL. They’d also built AI-agent and RAG projects. They listed Java, Python and C++ as their languages and were learning Kotlin.

At the time of the post, they were open to different teams and asking how to improve their chances of finding a match.

26 Recursion DP and a confusing email before team matching

Report date: Oct 13, 2026 · Interviews: Oct 5 · Role: Summer 2027 SWE Intern · Status: Team matching

A questionnaire arrived before the recruiter’s congratulations, then was recalled. The candidate briefly worried it had gone to the wrong person.

Their timeline:

• Aug 31: Applied after an early-batch rejection.

• Sep 5: Recruiter requested a transcript.

• Sep 6: Online assessment.

• Sep 13: Completed it.

• Sep 14: Application forwarded to Google’s recruiting team.

• Sep 27: Virtual interviews scheduled.

• Oct 5: Two back-to-back coding rounds.

• Oct 12 or 13: Team matching.

The first interviewer checked that they understood the question. It was an easy recursion problem with three follow-ups adding preliminary checks. There was still time for their questions.

The second round felt stressful: a medium dynamic-programming question, then an easier grid BFS follow-up. They were asked for time and space complexity and dry runs.

They’d practiced around 300 problems and recommended reading interview reports alongside coding practice. The graph follow-up felt familiar. They still credited luck, noting that friends with similar backgrounds hadn’t received an assessment.

The exact prompts weren’t shared.

27 Hints a bucket solution and a team matching update

Report date: Oct 13, 2026 · Interviews: Oct 6–7 · Role: U.S. MS SWE Intern, Summer 2027 · Status: Team matching

The candidate completed the assessment on September 10 and had two technical interviews on October 6 and 7.

Round one involved a graph problem and DFS. They discussed preprocessing, binary search, efficiency, edge cases, and other approaches.

Round two asked how to find the top trending videos. They started with a straightforward solution, improved it with a heap, then worked toward a bucket-based approach after the interviewer pushed for another optimization. In a reply, they confirmed they coded the bucket solution.

They also discussed dependency graphs, topological sorting, deployment failures, and recovery.

The candidate needed hints, made mistakes, and wasn’t completely happy with their implementation. They kept explaining their thinking and described both interviewers as professional and collaborative.

On October 13, they confirmed they had reached Project Search, also called team matching. They expected matching to begin in November, with a questionnaire and a possible host interview still ahead. They did not yet have an offer.

28 A 70 minute first round then a DP correction

Report date: Oct 10, 2026; updated Oct 12 · Role: U.S. bachelor’s SWE Intern, Summer 2027 · Result: Not reported

The first question connected to the candidate’s previous YouTube internship. The interviewer accepted the first solution, then changed the scenario and requested time and space optimizations. They joked that a VP, then an SVP, wanted further improvements. The candidate changed the code and walked through each case.

The schedule was two 45-minute rounds with a 30-minute break. Around minute 50, the interviewer offered to stop. They agreed to continue with a five-minute break. Round one lasted about 70 minutes. They felt the interviewer enjoyed the discussion.

Round two involved medium DP. They thought they solved it in five to ten minutes, but a dry run revealed a different interpretation of the prompt. They made a small correction, tested it again, and quickly improved the complexity.

This interviewer was quiet but relaxed. They also discussed their background and projects, with five minutes for questions.

29 A quick first round then a tree follow up they worked through together

Report date: Oct 10, 2026 · Role: U.S. SWE Intern · Result: Not reported

Round one was an unusual but basic combinatorics problem. The candidate solved it with about 15 minutes left. An easy follow-up took one extra line of code, leaving time to talk about Google.

Round two started with a tree problem. Again, they finished the main question with about 15 minutes remaining.

After a simple follow-up, the interviewer asked them to construct a tree with a property from the original question. The candidate pointed out a way to make that task trivial, so the interviewer added another restriction.

They worked through the final version together. With hints, they agreed on an approach that avoided brute force, though the candidate still wasn’t sure it was correct. They didn’t code that part. About five minutes remained, and the interviewer told them they had done well.

The exact tree property and added restriction weren’t shared.

30 A Latin square variation and an interviewer who kept pushing back

Report date: Oct 7, 2026 · Interviews: Oct 1 · Role: Summer 2027 SWE Intern · Status: Team matching

This was the candidate’s first time going through the full Google intern interview process.

• Sep 7: Applied.

• Sep 9: Recruiter reached out.

• Sep 14: Virtual interviews scheduled.

• Oct 1: Two back-to-back virtual interviews.

• Oct 7: Entered team matching.

There were no behavioral questions or project discussions. At most, there was a short introduction before coding.

The candidate described the questions as medium to hard. One was a Latin square variation; the other involved divide and conquer. They couldn’t find the original questions online.

They explained their approach before writing code. Depending on the interviewer, optimization happened either during implementation or after the first version was finished.

The first interviewer repeatedly questioned their approach, which made the round feel stressful.

For preparation, they worked through NeetCode 150, taking time to think through each problem. They felt it covered a wide range of data structures.

After reaching team matching, they were still unsure how that stage worked and asked others for advice.

31 Team matching with a Microsoft offer already accepted

Report date: Oct 7, 2026 · Interviews: Oct 5 · Role: Summer 2027 SDE PhD Intern · Status: Team matching

This candidate shared a short timeline:

• Aug 31: Applied with a referral.

• Sep 1: Received an interview invitation.

• Oct 5: Completed two coding interviews.

• Oct 7: Was told they had entered team matching.

They didn’t share the coding questions.

Their main concern was scheduling. They had already accepted a Microsoft offer and wanted to know whether Google might allow the internship to move to spring or winter.

In a reply, they added that they expected to graduate the following year.

The post didn’t include an answer from Google about changing seasons. At that point, they were in team matching.

32 A tree question a small fix and a harder second round

Report date: Oct 7, 2026 · Role: U.S. SWE Intern · Result: Not reported

After waiting about two weeks, the candidate completed two technical rounds.

The first question involved a tree. They finished in about 20 minutes, then handled two follow-ups with small code changes. One asked what they would do before handing the code to a manager.

During a dry run, the interviewer found a small error. The candidate said it came from a misunderstanding and fixed it.

The second question felt much harder, close to LeetCode hard. Previous practice with a similar problem helped them find the optimal approach quickly. The interviewer told them to start coding. They forgot to discuss complexity first, and the interviewer didn’t ask for it.

The dry run went well. They explained one follow-up and how they would implement it, but the interviewer didn’t require more code.

Both interviewers were friendly and left time for questions. The candidate hadn’t received a team-matching decision when they posted.

33 An OOP question given aloud with three follow ups

Report date: Oct 6, 2026 · Role: U.S. BS SWE Intern, 2027 · Result: Not reported

This candidate described their first technical round. The question involved OOP and data structures and felt around LeetCode medium difficulty.

They struggled at first. They hadn’t used OOP for about a year, and the interviewer explained the whole question aloud without pasting the prompt.

Once they understood it, they wrote pseudocode, explained their thinking, and started coding. They worked through the main question and three follow-ups, discussing edge cases and time and space complexity along the way.

The interviewer checked the code repeatedly and accepted sections before moving on. The candidate was nervous and felt their explanation could have been better organized.

A small control-flow bug remained at the end of the last follow-up. The interviewer gave a hint, but after about three minutes they ran out of time without fixing it.

Only the first round was described, with no recruiter decision.

34 Greedy then backtracking after a recursion mistake

Report date: Oct 3, 2026 · Role: U.S. SWE Intern, 2027 · Result: Not reported

A scheduling error left four days between the rounds. Both were coding-only, with no follow-ups.

The first question involved greedy algorithms, which they hadn’t studied. They explained an approach, got the interviewer’s approval, coded it, and walked through a test case. They believed it worked but wasn’t fully optimized, and explained its time and space complexity. The interviewer seemed relaxed, though little time remained for questions.

Round two became a backtracking problem. A discussion about memory and operation space caused confusion; they thought they and the interviewer were interpreting it differently.

The interviewer pointed them toward recursion. Their first recursive implementation was wrong. As they caught mistakes, they recognized the need for backtracking and eventually reached a solution.

They felt the interviewer became excited as their reasoning developed. The round ran over as they asked technical questions about Google.

They had applied with a mentor’s referral and had one previous internship. They hadn’t practiced LeetCode for about a year and had only moved past two pointers two weeks earlier.

35 No response to the early application then a quick move to Project Search

Report date: Sep 30, 2026; edited Oct 2 · Interviews: Sep 29 · Role: Summer 2027 SWE Intern · Status: Project Search

Google was the first company where this candidate had made it through the interviews without being rejected. They sounded relieved, then immediately worried about the next stage.

Their timeline included two applications:

• Jul 21: Applied to the early batch using a personal email, without a referral. No follow-up.

• Sep 13: Applied to the regular batch using a school email, this time with a referral.

• Sep 15: Online assessment.

• Sep 21: Interview invitation.

• Sep 29: Two back-to-back coding rounds.

• Sep 30: Entered Project Search.

They felt lucky because both interview questions seemed easy. They didn’t share the actual prompts.

After seeing other candidates post their backgrounds online, they worried about failing to find a match. They asked whether they should actively reach out to teams and how others had handled Project Search.

36 A resume deep dive then First Missing Positive

Report date: Sep 20, 2026 · Role: Google Intern

The interview started on time. The first ten minutes were an intense resume discussion, focused on the candidate’s recent backend work.

The interviewer asked detailed questions about the Kafka streaming architecture and Redis caching they had worked on during a previous internship in the Platform Cloud org.

They also discussed Bastion, including the technical trade-offs, why the candidate chose Go, and how they handled latency in a distributed edge API gateway.

Coding: First Missing Positive — LeetCode 41

The rest of the interview focused on First Missing Positive.

The candidate first proposed a hash set solution with O(N) time. The interviewer then asked for O(1) extra space.

They switched to an in-place cyclic-sort approach: move each valid number x to index x − 1, then look for the first missing positive.

They wrote the solution in Python and handled negatives, out-of-range values, and duplicates. The duplicate check prevented repeated swaps of the same values.

In the final few minutes, they walked through edge cases supplied by the interviewer, followed by a short Q&A about the team.

37 Three coding questions across two 45 minute rounds

Report date: Sep 5, 2026 · Interviews: Aug 28 · Role: Summer 2027 SDE Intern · Status: Team matching

This candidate applied through the early-consideration batch.

• Jul 20: Applied with a referral.

• Aug 13: Received the online assessment.

• Aug 17: Received the interview-scheduling email.

• Aug 28: Two back-to-back virtual interviews, 45 minutes each.

• Aug 31: Received an email saying they had passed the interviews and entered team matching.

Round 1

There were two questions: a DFS/backtracking problem and a variation of Task Scheduler.

Round 2

There was one BFS path-finding question.

The candidate confirmed in the comments that both rounds were coding-focused, with no behavioral questions. They also clarified that the interviews were virtual.

The post didn’t include enough detail to identify the exact problems. After entering Project Search, they asked people who had been through it for team-matching advice.

Earlier reports: Part 1 · Part 2 · Part 3