← Back to Directory

CloudKitchens interview notes, page 2 of 2

Mid-sized

Screna has 12 CloudKitchens interview write-ups. The most reported round is the technical phone screen (5).

12 interview notes · updated Jul 2026

CloudKitchens·Software Engineer·Online Assessment (OA)

Jul 2026
CloudKitchens SWE interview was a multi-level OA where you build out an in-memory cloud storage system from scratch, adding features layer by layer. The problem itself is pretty involved and covers file management, user quotas, merging, backups, and size transforms. No system design fluff, just pure implementation.
  • Design and implement an in-memory cloud storage system with basic file operations: adding a file (failing on duplicate names), retrieving a file's size, and deleting a file.
  • Extend the storage system to support per-user file ownership and a query that returns the top N files for a given user, filtered by name prefix and suffix, sorted by size descending with lexicographic name as a tiebreaker.
  • Add user creation with capacity limits, a file-add operation that enforces per-user quotas, a capacity adjustment that auto-deletes the user's largest files if needed, and a user merge that transfers files while respecting the target user's capacity.
  • Implement backup and restore for user files, plus compress (halve each file size) and decompress (double each file size, then enforce capacity by deleting largest files) operations.

“This part is straightforward.”

View Post

CloudKitchens·Software Engineer·Onsite - Coding / Algorithms

Jul 2026
CloudKitchens SWE interview that was basically one big problem stretched into three increasingly painful variants. They start you off with something that feels manageable and then just keep adding constraints until you're questioning your life choices.
  • Given an array of integers and a value k, return the k most frequent elements. How do you handle ties in frequency?
  • Design a data structure that supports adding integers from an unbounded stream and querying the top-k most frequent values at any point.
  • Extend the streaming design to only consider events within the last 3600 seconds. Events arrive as (timestamp, value) pairs with monotonically non-decreasing timestamps. Support add and topk operations efficiently.

“This part felt like a warmup and I treated it that way, which was maybe a mistake because they immediately pushed on tie-breaking and I hadn't thought about it at all.”

View Post