RepostedGoogle logo

Google · Phone screen → Onsite

In process

Google Intern Coding and OOP: 19 Experiences (Part 5 of 8)

Software engineer · InternshipBachelor's degreeOct 2026
Shared by

Google SWE intern reports from Oct 2026: two 45-minute coding rounds, graphs and trees in 9 of 19, then DP, grids, prefix sums and object-oriented tasks.

At a glance

  • Process: sometimes a 15-minute resume screen, then two 45-minute coding rounds, often back to back on one Meet link with a shared doc and no code execution (45, 51, 52, 55, 56). Some candidates were asked for a third interview (40, 43); others went to team matching (50, 52, 54).
  • Asked recently: graphs or trees in 9 of 19 reports, DP or bitmasking in 4, grids in 3, prefix sums in 1 (two rounds), object-oriented design in 3. Named tasks include eight-direction islands (40), maximum-sum subarrays with equal endpoints (51) and rental-car assignment (53).
  • Difficulty: many finished a working but not optimal answer, and the request for O(1) space or a faster bound was where rounds slipped (38, 39, 45, 49). Two OOP rounds were the hardest described (44, 47).
  • Practise first: Number of Islands with Diagonal Connectivity, Maximum Subarray Sum with Equal Endpoints and Maximum Subarray Endpoint Score.

Coverage: Google SWE / SDE intern, BS, MS and PhD, U.S. and Canada where stated. Nineteen reports, entries 38–56, posted from August 30 to October 21, 2026. † marks reports that did not confirm the Summer 2027 cycle.

In this series

  • Part 1: graphs, intervals and OOD from the first wave of 2027 intern loops.
  • Part 2: paths, grids and trees.
  • Part 3: prefix sums and heaps.
  • Part 4: full loops and team matching.
  • Part 5 (current): grids, DP, OOP and AI.
  • Part 6: trees and knapsack.
  • Part 7: PhD and return offers.
  • Part 8: updates and host calls.

What came up

Graphs and trees still lead, but this batch adds object-oriented rounds and two back-to-back prefix-sum questions.

Topic Reports Last seen
Graphs and trees 9 of 19 Oct 2026 (report date)
DP or bitmasking 4 of 19 Oct 2026 (report date)
Grids and boards 3 of 19 Oct 2026 (report date)
Object-oriented design 3 of 19 Oct 2026 (report date)
Arrays, prefix sums or sliding window 3 of 19 Oct 2026
Caches, queues or heaps 3 of 19 Oct 2026 (report date)
Strings and formatting 2 of 19 Oct 2026 (report date)
Resume, projects or AI workflow 2 of 19 Sep 2026 (report date)
Filesystem 1 of 19 Sep 2026

A candidate can appear in more than one row.

Questions we found

Algorithms / Data Structures

Question Reports Last seen
Number of islands, eight directions 1 Oct 2026 (report date)
Water cells next to land (practice reconstruction) 1 Oct 2026 (report date)
2D game-board validation (practice) 1 Oct 2026 (report date)
LRU cache plus a graph (cache practice) 1 Oct 2026 (report date)
Text Justification-like formatting 1 Oct 2026 (report date)
Longest Duplicate Substring-like (LeetCode 1044) 1 Oct 2026 (report date)
Maximum subarray sum with equal endpoints 1 Oct 2026
Endpoints minus the middle 1 Oct 2026
Add and remove on a custom set 1 Oct 2026
Rental-car assignment 1 Oct 2026
Count visible people 1 Jun 2026

Resume, projects and AI workflow

Question Reports Last seen
Resume and basic profile 1 Sep 2026 (report date)
How you use AI in your workflow 1 Sep 2026 (report date)
Past projects, DSA practice and coding-contest activity 1 Jun 2026

19 detailed candidate timelines, loops, and screens

38: LRU and graph hybrid, then bitmask DP †

Report date: Oct 21, 2026 · Role: SWE Intern · Result: Pending

  1. Round 1: a mix of LRU cache and a medium graph problem, solved with a few hints. The LRU practice covers only the cache part.
  2. Round 2: much harder. They partly coded an O(4^n) approach, later hearing of an O(n × 2^n) bitmask DP. They felt the interviewer did not expect the optimum.

39: A Longest Duplicate Substring–like question †

Report date: Oct 21, 2026 · Role: SWE Intern (from the intern thread)

  1. Round 2 was a one-line question they solved non-optimally. The interviewer kept asking for better; they proposed a hashmap and ran out of time. They later found it close to Longest Duplicate Substring (LeetCode 1044). No result by Oct 21.

40: Eight-direction islands and adjacent water

Report date: Oct 20, 2026 · Role: Summer 2027 SWE Intern · Status: Third interview requested

  1. Round 1 (45 min): count islands with DFS, first checking four neighbors and then correcting to eight; complexity misstated. Follow-up: water cells next to land; a missing if hurt the code. The practice version's exact rules are authored.
  2. Round 2 (45 min): optimal solution, main task done about 20 minutes early, three follow-ups answered, 15 minutes of chat. They forgot a deque import. On Oct 21 a third interview was requested.

41: A Text Justification–like question †

Report date: Oct 20, 2026 · Role: SWE Intern

  1. Round 1 (45 min): their own approach, minus a hashmap the interviewer questioned; main task and follow-up coded and dry-run.
  2. Round 2 (45 min): a one-sentence prompt that took 10–15 minutes to clarify. The approach was accepted but a confirmed edge case was left unhandled. They compared it to Text Justification.

42: The platform failed in round two

Report date: Oct 20, 2026 · Role: 2027 SWE Intern

  1. Round 1 went fine. In round 2 the coding platform stopped accepting input, and the interviewer said to reschedule. Questions were not shared.

43: Queue logic and board validation

Report date: Oct 19, 2026 · Role: Summer 2027 SWE Intern · Status: Third interview invited

  1. Round 1: a control-flow and queue task needing long clarification; core logic present, code unfinished, the round 5–10 minutes over.
  2. Round 2: validate a game board in a 2D array. Graphs were unnecessary, said the interviewer; they discussed checks such as boundaries and DFS for some cases but did not finish. Validate Rectangular Game Pieces is a related reconstruction.

44: A Canadian intern's OOP round †

Report date: Oct 19, 2026 · Role: SWE Intern · Location: Canada

  1. Two interviews; they struggled with an object-oriented design task and would have preferred graphs or trees. The system, the other round and the result were not shared.

45: PhD intern: grid DP and a tree

Report date: Oct 16, 2026 · Role: Summer 2027 SWE PhD Intern · Result: Pending

  1. An alumnus referral; about six days to prepare. Two 45-minute rounds 15 minutes apart on one Meet link, writing Python in Google Docs with no execution or highlighting.
  2. Round 1: an m × n grid DP. Brute force first; asked for an O(1) improvement, they proposed an auxiliary-array DP but did not finish.
  3. Round 2: an unusual tree task with recursive DFS; initial and follow-up code worked on their own example. Follow-ups mentioned: an O(1) optimization, and converting iterative code to recursive while keeping a cache.

46: BS intern: bitmasking and a rejection

Report date: Oct 16, 2026 (details Oct 21) · Role: BS SWE Intern, 2027 · Result: Rejected

  1. Round 1: an unclear bitmasking task; they did not finish its first part. Three follow-ups went unreached.
  2. Round 2: better, but a missed edge case stayed unfixed.

47: MS intern: designing an internal tool †

Report date: Oct 15, 2026 · Role: MS SWE Intern · Location: U.S.

  1. Round 1: object-oriented design for an unnamed internal Google tool, over an hour. With many hints they built part of it; the interviewer said nobody had fully finished it.
  2. Round 2: wordy, but a simple DFS, done in about ten minutes, then follow-ups on alternatives, flaws and complexity.

48: A working DFS near time †

Report date: Oct 15, 2026 · Role: SWE Intern · Location: U.S.

  1. Round 1: a medium-to-hard graph task; a working, non-optimal recursive DFS near the deadline. The interviewer valued working code and liked the explanation.
  2. Round 2: a medium task: brute force, then their own optimization while coding, a follow-up and complexity questions.

49: The tree solution needed constant space

Report date: Oct 15, 2026 · Role: Summer 2027 SDE Intern · Location: U.S.

  1. Round 1: graph DFS, clarified, explained, coded and dry-run; about 20 minutes left for conversation.
  2. Round 2: a tree task with optimal time but O(N) space; asked for O(1), they could not find it in time.

50: A neighbor hint, then team matching †

Report date: Oct 14, 2026 · Role: SWE Intern · Status: Team matching

  1. Round 1: a hint about checking neighbors unlocked the main problem; follow-ups solved, one non-optimally.
  2. Round 2: main task and follow-ups done in about 15 minutes. A later edit confirmed team matching.

51: Two rounds, both prefix sums

Interview date: Oct 13, 2026 · Role: U.S. Summer 2027 SWE Intern · Result: Unsure

  1. Round 1: a quiet interviewer pasted maximum-sum subarray with equal endpoints: pick i ≤ j with nums[i] == nums[j] to maximize the sum from i to j; return the indices. A single element counts. They went from O(n³) to prefix sums plus a map of the smallest prefix before each value, O(n). On [1,2,3,2,5] the answer was [1,3], sum 7. The follow-up dropped the equal-endpoint rule (one global minimum prefix); they misread a request to keep the original function, explained it without finishing, and corrected space from O(n) to O(1) after a prompt.
  2. Round 2: endpoints minus the middle: a subarray of at least two elements scores its first plus last value minus everything between; return the best. They split the score into left and right parts and kept the best left contribution in one pass. Two bugs got fixed: starting at 2 * nums[0] (a one-element interval) and prefix initialization out of order. O(n) time, O(1) space; they explained a cleanup instead of rewriting.

52: Canada: team matching

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

  1. Two 45-minute rounds, each with clarification, brute-force pseudocode and complexity, then optimized Python.
  2. Round 1: code done, but no dry run or follow-ups. Round 2: code and dry run done; a hint fixed a pointer-tracking slip; several follow-ups. They later called the graph and DP questions medium-hard, then hard.
  3. Team matching confirmed Oct 17.

53: A custom set and rental cars

Report date: Oct 9, 2026 · Role: SWE Intern

  1. Round 1: implement add and remove on a special set the interviewer described. About 30–40 minutes to understand and code it with almost no hints; a small issue fixed while testing. No follow-up.
  2. Round 2: the rental-car assignment problem: minimum cars and the ID assignments. They went straight to a min-heap, finished early, then wrote and walked through test cases. The leftover time felt awkward. No result shared.

54: PhD intern waiting for a team

Interview date: Sep 24, 2026 · Report date: Oct 20, 2026 · Role: Summer 2027 SWE PhD Intern · Status: Team matching

  1. HR contact Sep 14; two back-to-back technicals Sep 24; team matching Sep 29.
  2. LeetCode-style coding with several follow-ups, no research questions. One was a filesystem question. By Oct 20 the questionnaire was done and matching was pending; they sought LLM, agents or ML teams.

55: An AI-workflow discussion

Report date: about Sep 2026 · Role: 2027 SWE Intern

  1. A 15-minute resume and profile screen, then two 45-minute elimination rounds.
  2. Round 1: an array / sliding-window task; they wished they had followed a hint about min. A discussion on how they use AI in their workflow followed; there was no AI coding task.
  3. Round 2: a graph task, then resume talk.

56: Visibility for one person, then everyone

Interview date: Jun 29, 2026 · Report date: Aug 30, 2026 · Role: Summer 2027 SDE Intern · Result: Rejected

  1. A strong coding-contest result led to recruiter contact. A 15-minute screen covered projects, DSA practice, problem counts and contest activity.
  2. Technical: people of distinct heights face one way; count whom the last person can see, where no one between is taller than both. An optimal O(n) scan; the follow-up for everyone's count needed a stack, which they did not finish. They likened it to LeetCode 1944, whose rule differs slightly.
  3. Two rounds were planned; they were rejected after one.

Patterns across the reports

  1. The space follow-up. Requests for O(1) space or a faster bound stopped 39, 45 and 49, and pushed 38 toward bitmask DP.
  2. OOP rounds ran long and hard. 44 and 47 called their object-oriented tasks the hardest part; 47's ran over an hour.
  3. Third interviews. Two candidates were asked for a third round after mixed results (40, 43).

Common questions

Does Google ask object-oriented design to interns?

Sometimes: three reports had an OOP round, including an internal-tool design nobody had finished (44, 47, 53).

Can you run code in a Google intern interview?

Not in 45's loop: Python in Google Docs with no execution or highlighting. 56 was briefed on a shared sheet and Meet.

Why would Google ask for a third intern interview?

Both candidates asked had one round go well and one unfinished (40, 43).

Does Google ask about AI use in intern interviews?

One candidate discussed how they use AI in their workflow, with no AI coding task (55).

Keep practicing

In this series

Sources: Anywhere useful interview info can be found.

0 total reactions. Your reactions: none.
0 comments

Comments

0

No comments yet — be the first to share what you know.