RepostedConfluent logo

Confluent · Phone screen → Onsite

In process

Confluent SWE OA, Coding and Design Interviews: 19 Experiences

Software engineer · UnknownAug 2026
Shared byjanebeingjane

Confluent SWE interviews, Apr 2024–Oct 2026: 19 reports, led by temporary-email system design (11) and the sensor-health coding task (6).

At a glance

  • Process: a HackerRank OA or a phone screen, then an onsite of coding, system design and a values round (02, 05, 17). Report 10 said three rounds were the full loop for its level.
  • Asked recently: temporary-email design for 100 million users in 11 of 19 reports, sensor health in 6, word and phrase search across documents in 6, and variadic function matching in 4.
  • Difficulty: design decided many outcomes: 07, 11 and 14 passed coding but failed design; 15 got an offer.
  • Practise first: Durable disposable email registry, Silent Sensor Detector and Sequential Phrase Search Across Documents.

Coverage: 19 Confluent software engineering candidates, including Senior I and senior roles where stated; OA, phone screens, onsite coding and design; India and Canada where stated; reports April 2024 to October 2026. The file-reading tasks from these loops are in the file manipulation round.

In this Confluent collection: question collection · file manipulation round · this supplement · concurrency, part 1 · concurrency, part 2

What came up

Topic Reports Last seen
Named coding or search task 18 of 19 Oct 2026
System design 14 of 19 Oct 2026
Behavioral or project discussion 7 of 19 Oct 2026

Questions we found

Dates are report dates unless marked.

Coding

Question Reports Last seen
Sensor health with wasAlive 6 Sep 2026 (interview)
Word search across documents with a phrase follow-up 6 Oct 2026
Variadic function matching 4 Oct 2026
Sudoku Board Validation 2 Oct 2026
Sudoku Puzzle Solver 2 Oct 2026
Windowed key-value map (expiring cache) 2 May 2024
Count Combinations Using Weights One Through K 1 Oct 2026
Sum Symmetric Pairwise Remainders 1 Oct 2026
Combination sum with negative numbers 1 Oct 2026
Monster fighting in two parts (part 1) 1 Sep 2025
Time-window key-value structure (an LRU variation) 1 Apr 2024

Design

Question Reports Last seen
Temporary email service 11 Sep 2026 (interview)
Random queue (low-level design) 2 Aug 2026
RSS feed aggregator 2 May 2026
TinyURL 2 May 2024
Podcast subscription feed 1 Oct 2026

Behavioral (BQ)

Question Reports Last seen
Why Confluent? 1 Oct 2026
Discuss one past project in depth 1 Oct 2026
A DevProd role after asking for distributed systems 1 Sep 2026
From customer-facing teams to DevProd 1 Sep 2026
Project phases and handling situations 1 May 2024
Project with the greatest impact, and its biggest challenge 1 Apr 2024

19 detailed candidate timelines, loops, and screens

All dates are report dates unless marked.

01: Two OA counting questions

Report date: October 14, 2026 · Stage: 75-minute coding assessment, camera off

  1. Count weight combinations: use weights 1 through k to reach total, return the count modulo 1,000,000,007 (for total = 8, k = 2, the answer is 5). Only the count was needed; the candidate compared it to coin-change counting.
  2. Sum pairwise remainders: for all 1 <= i, j <= n, sum i % j + j % i modulo 1,000,000,007, with n <= 10^6 (n = 2 gives 2). The candidate summed full remainder cycles plus a partial one and used symmetry.

02: A hashmap hint, loose phrases and AND/OR queries

Report date: October 13, 2026 · Level: Senior I · Location: Canada

  1. Phone screen: variadic function matching. The candidate had prepared a trie; the interviewer steered them to a hashmap, and they finished.
  2. Onsite coding 1: exact word match across documents, then a phrase whose words need not be adjacent, then queries joining words with AND or OR, which felt improvised after they finished the first two fast.
  3. Onsite coding 2: Sudoku validation and solving by backtracking.
  4. System design: a podcast publishing system, built from fan-out on read to fan-out on write to a hybrid.
  5. Values: why Confluent and a deep dive into one project.

03: Combination sum, from yes/no to every answer

Report date: October 8, 2026 · Stage: onsite coding

  1. Does a combination exist with all-positive numbers? Then with positive and negative numbers? Then return one solution, then all, then optimize. Reuse rules and length limits were not given; our version states them.

04: Sensor health and a 100-million-user email service

Interview date: September 29, 2026 · Status: awaiting a decision

  1. Coding: sensor health with wasAlive.
  2. Design: a temporary email service for 100 million users. Username generation took most of the time, leaving little for storage; the candidate proposed a Bloom filter for duplicate checks.

05: Sudoku on the phone and questions about DevProd

Report date: September 28, 2026 · Role: Software Engineer

  1. OA: 60 minutes on HackerRank, three unnamed coding questions.
  2. Phone DSA: validate a Sudoku board and solve Sudoku (report 02 had these onsite).
  3. Machine coding: find a word across documents, then a phrase with its words in order.
  4. Values, with the hiring manager: you wanted distributed systems; how do you feel about DevProd? and after customer-facing teams, can you handle DevProd, and how?

06: Binary search for sensor health

Report date: September 24, 2026 · Status: awaiting a decision

  1. Sensor health: brute force first, then binary search for wasAlive(key, start) over a window. Both coding rounds passed.
  2. Temporary email design: 100 million users, realistic usernames, fast duplicate checks.

07: Random queue, then two failed design rounds

Report date: August 28, 2026 · Result: Rejected

  1. A random queue with a comparison follow-up the candidate did not finish.
  2. Temporary email design: 100 million users, realistic names, fast de-duplication; the interviewer suggested Bloom filters. A second design round was not described.
  3. The recruiter said both coding rounds passed and both design rounds failed.

08: Word and phrase search, then email persistence

Report date: August 12, 2026 · Result: Rejected

  1. Word search with a phrase-search extension; positive coding feedback.
  2. Temporary email for 100 million users: Bloom filter vs hash lookup, and persistence. The design result was negative.

09: Variadic signatures, sensor health and temporary email

Report date: August 8, 2026 · Result: Rejected

  1. Phone screen: a variadic function registry matching signatures to argument types.
  2. Onsite coding: did a sensor stay alive during a window?
  3. Design: temporary email for 100 million users with plausible names and fast duplicate checks.

10: Variadic functions and duplicate-name checks

Report date: August 6, 2026 · Result: Rejected

  1. Phone screen: variadic function design.
  2. Temporary email design focused on duplicate names at scale; the candidate struggled to explain Bloom-filter trade-offs.
  3. Three rounds were the whole loop for their level; the rejection came two days later.

11: Phrase search and two design rounds

Report date: June 15, 2026 · Result: Rejected

  1. Coding: phrase search, interface not given.
  2. Design 1: temporary email (Bloom filters, APIs, scale), ending with about 15 minutes spare. Design 2 was not named.
  3. Coding passed; both design rounds failed.

12: Bloom filter before the database

Report date: June 6, 2026 · Role: senior software engineering · Status: awaiting a decision

  1. Sensor health: key/timestamp pings and wasAlive(key, start), with binary search added after the base version.
  2. Temporary email: 100 million active users, human-looking addresses, high volume; a Bloom filter in front of a definitive database lookup.
  3. Behavioral questions were mentioned but not shared.

13: Random queue in a virtual onsite

Report date: May 23, 2026 · Result: Rejected

  1. Random queue design; the interface was not given. The rejection came two days later.

14: Sensor monitoring, RSS and email persistence

Report date: May 17, 2026 · Result: Rejected

  1. Coding: sensor-health monitoring, which felt acceptable.
  2. Design: an RSS feed aggregator, then temporary email (usernames, Bloom filters, scale, persistence). The recruiter confirmed both design rounds failed.

15: Three inactive slots and an RSS API, then an offer

Report date: May 7, 2026 · Result: Offer

  1. wasAlive(sensorID, startTime): three inactive 100-millisecond slots in a row mean the sensor is down. The candidate added binary search; the full window length was not given.
  2. Design: an RSS aggregator (REST/RPC, mobile and web), then temporary email for 100 million users with a Bloom filter before the database.
  3. The offer came two weeks after the onsite.

16: A two-part monster problem and variadic functions

Report date: September 3, 2025 · Level: Senior · Status: hiring committee

  1. Phone screen: variadic functions.
  2. Onsite: a two-part monster task. Part 1 is Find Defenders Against Every Hostile Monster (solved with DFS); part 2 was harder and its rules are unknown.
  3. Design: temporary email. Behavioral was "standard". Positive recruiter feedback before committee.

17: An expiring cache and shared search preprocessing

Interview date: April 2025 · Report date: July 18, 2025 · Role: SSE (1.5 years' experience) · Location: India · Result: offer paused by a hiring freeze

  1. Round 1: a cache with expiry: add timestamped pairs, read them, average active values; the candidate raised timestamp order, memory and read- vs write-heavy load.
  2. Round 3: return document IDs containing a word, then one preprocessing step for word and phrase search; proposed but not finished.
  3. Values: projects, challenges, volunteering, mentoring and teamwork. Passed values and team match before the freeze.

18: Runnable code and a project deep dive

Report date: May 16, 2024 · Role: Senior Software Engineer · Result: Offer

  1. Qualifier on CoderPad, about 20 days after referral: a windowed key-value map with complexity and locking follow-ups.
  2. Onsite coding: word search across documents plus a sequential phrase search; both needed running code.
  3. TinyURL design and a project and culture deep dive, including project phases. The candidate followed up twice before the hire decision and a role-sell call.

19: A time-window map, document search and TinyURL

Report date: April 19, 2024 · Result: Rejected

  1. Phone screen: a key-value structure over a time window, described as an LRU variation.
  2. Onsite: a word-search engine returning document IDs, then phrase search; then TinyURL design; last, the project with the greatest impact and its biggest challenge.

Patterns across the reports

  1. Temporary email decides many loops. 11 reports had it; 07, 11 and 14 passed coding but failed design, while 12 and 15 used a Bloom filter before the database and 15 got an offer.
  2. Sensor health moves to binary search. Reports 06, 12 and 15 added binary search after a brute-force version.
  3. Document search grows. Word search turned into phrase search in 02, 05, 08, 17 and 19; 02 added AND/OR queries.

Common questions

What is the Confluent system design interview?

Most often a temporary email service for 100 million users (11 reports), focused on realistic usernames and fast duplicate checks.

What coding questions does Confluent ask?

Sensor health with wasAlive (6 reports), word and phrase search across documents (6) and variadic function matching (4).

How many rounds is the Confluent onsite?

Report 10 said three rounds were the full loop; others list two coding rounds, one or two design rounds and a values round (02, 07, 14).

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.