RepostedGoogle logo

Google · Phone screen → Onsite

Rejected

Google Experienced SWE First Round and Onsite: 1 Experience

Software engineer · Full-timeMaster's degreeReported Nov 13, 2026
Shared byjanebeingjane

A Google experienced SWE in 2026 passed a Course Schedule first round, then had a Mountain View onsite with a sliding window and a swipe-keyboard trie.

At a glance

Coverage: One experienced software engineer with a Master's degree, changing jobs, referred to Google by a friend. The report covers a first round in May or June 2026 (coding and behavioral), a delay of two to three months, a two-round coding onsite in Mountain View, and the hiring committee decision. It gives October to December 2026 as the interview period and went up on November 13, 2026.

What came up

All three coding questions were graph, sliding-window or trie work close to well-known problems, and the candidate needed a hint in both onsite rounds. The swipe-keyboard round went deepest, with follow-ups on skipped letters, thread safety and duplicate results.

Topic Reports Last seen
Swipe-keyboard word finder with a trie and DFS 1 of 1 Nov 2026 (report date)
Longest subarray whose values stay within a limit (sliding window) 1 of 1 Nov 2026 (report date)
Hiring committee rejection after positive recruiter feedback 1 of 1 Nov 2026 (report date)
Course Schedule variant 1 of 1 Nov 2026 (report date)

Questions we found

Algorithms / Data Structures

Question Reports Last seen
Swipe-keyboard word finder: given the letters a finger passed over and a word list, find the words that can be typed (Batched Subsequence Word Matches) 1 Nov 2026 (report date)
Longest run of consecutive values whose spread stays within a limit, similar to Longest Continuous Subarray With Absolute Diff Less Than or Equal to Limit (Longest Subarray Within an Absolute-Difference Limit) 1 Nov 2026 (report date)
Course Schedule variant (Course Schedule Cycle Detection) 1 Nov 2026 (report date)

1 detailed candidate timeline

01: Experienced SWE, a long process ending at the hiring committee

Interview period: May–June 2026 (first round) and 2026 Q4 (onsite) · Report date: November 13, 2026 · Role: Software engineer · Degree listed: Master's · Channel: Referral · Candidate's rating: Negative, average difficulty · Result: Not selected

First round (May–June 2026)

The process started about six months before the report, through a friend's referral, and the recruiter moved the candidate to a different role once along the way.

  1. Coding. A variant of Course Schedule. The candidate knew this kind of question well, solved it smoothly, handled the follow-up and talked through many test cases with the interviewer. The report doesn't say how the variant differed. The practice version, Course Schedule Cycle Detection, asks whether every course can be finished.
  2. Behavioral. This also went smoothly. The report doesn't list the questions.

The recruiter called the performance impressive and moved the candidate to the onsite.

The delay

Scheduling then stalled for two to three months. The person handling the immigration check couldn't reconcile the candidate's nickname with their legal name, and the candidate wasn't sure their several emails explaining it were even read. In the end a different recruiter set up the onsite, in Mountain View. The candidate flew in from the East Coast and had slept badly two nights in a row, which left them a bit foggy on the day. Both interviewers were friendly and easy to talk to.

Onsite round 1: longest subarray within a limit

  1. The first 10 to 15 minutes were about the candidate's work history and projects they found interesting.
  2. The question was close to the well-known problem of finding the longest run of consecutive array values whose largest and smallest differ by at most a limit. The candidate proposed two pointers with a sliding window, and the interviewer agreed.
  3. While coding, the candidate remembered that the problem can also use a deque, added a stack on top of it and got tangled for two or three minutes. The interviewer pointed out that the stack wasn't needed, and the candidate straightened out the code and finished it.
  4. There was a follow-up coding part, which the candidate also completed. The report doesn't describe it.
  5. Practice version: Longest Subarray Within an Absolute-Difference Limit. It uses the standard form of the problem; the follow-up isn't included.

Onsite round 2: swipe-keyboard word finder

  1. This round went straight to the question, which the candidate compares to Word Search II. The setting is swipe typing on a phone keyboard: from the letters a finger slides over, find the words in a word list. At first the interviewer only demonstrated on a phone without defining the input, and the candidate briefly thought a tokenizer was wanted. Then came an example: the swipe "cxsasert" with the word list cat, sat, you and insert should give cat and sat.
  2. The candidate offered a trie or a BFS scan. The interviewer liked the trie, so the candidate built the word list into a trie once, with each node holding its children and an end-of-word flag, so later lookups could reuse it.
  3. The candidate raised an edge case on their own: matching a prefix isn't enough, the search has to keep going deeper. They wrote a DFS in the style of Word Search II that tracked the current trie node, the position in the swipe string and the letters collected so far. The interviewer pointed out two typos, which the candidate fixed.
  4. Follow-up: skipped letters. The interviewer asked how a swipe like "gokgle" should still find "google". The candidate first thought of regular expressions but dropped the idea. With a hint, they changed the search so that each letter has two options: follow it in the trie if it matches a child, or skip it as a letter the finger merely passed over. Both options move one step forward in the swipe string. The report doesn't explain how this rule produces the doubled letter in that example.
  5. The interviewer asked for the candidate's thinking every few steps. The candidate liked that style, but it made the coding slow, and there was no time to code the last follow-up; the interviewer said explaining the idea was enough.
  6. Follow-up: thread safety. The candidate had stored results in an instance field. The interviewer pointed out this breaks with multiple threads, so the candidate changed the function to return the result list.
  7. Follow-up: weaknesses. Asked what was wrong with the approach, the candidate first thought of unrelated words being matched. The interviewer meant that the same word can be found several times through different skip paths, giving duplicates. After asking two or three times to clarify, the candidate used a set to remove them.
  8. Practice version: Batched Subsequence Word Matches returns the words whose letters appear in order in the swipe string. It judges only the returned words, not the trie, thread-safety or duplicate discussion.

Result

The candidate left feeling uneasy, since both onsite rounds had needed hints. A week later the recruiter reported positive feedback and handed the candidate to another recruiter. In a call with the new recruiter, the candidate heard that the feedback praised code comprehension and data structures and algorithms in the coding rounds, and that the behavioral round was also good. The recruiter then asked for an updated résumé to send to the hiring committee.

Then came almost four weeks of waiting, with few replies to the candidate's emails. When they finally spoke, the answer was a rejection because coding performance was mixed. The candidate asked about an extra round or a lower level and was told the decision was final, with no such options, and a 12-month cooldown.


Common questions

What is the Google swipe-keyboard interview question?

Given the letters a finger slid over and a word list, find the words that could have been typed ("cxsasert" with cat, sat, you, insert gives cat and sat), then handle skipped letters, thread safety and duplicates (report 01).

Can positive recruiter feedback still end in a Google rejection?

Yes, in this report: the recruiter called the feedback positive, but the hiring committee rejected for mixed coding, with a 12-month cooldown and no extra round or lower level offered (report 01).

Keep practicing

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.