Glean Search

Role: Software Engineer

College: Not Specified

Interview Date: January 3, 2023

Interview Type: Full-Time

Result: Not Specified

Difficulty: Hard

Rounds: At least 1 coding round

Topics Asked: Algorithms, Data Structures (queues, array manipulation), Concurrency/Scheduling concepts

Experience:

A coding interview for a Software Engineer position at Glean involved a detailed problem focused on document processing and indexer allocation.

• Coding Round: The problem described a system where "Documents are indexed using a processing queue. They are queued at different timestamps and processed using parallel indexing jobs running across m indexers numbered 0 to m-1. Each indexer is busy when processing a document and cannot process any other documents during that time."

  Document `i` is initially assigned to indexer `i%m` if it's not busy. If the assigned indexer is busy, the system finds the next available indexer in increasing order (wrapping around from `m-1` to 0). If all indexers are busy, the document is dropped.

  The candidate was provided with:

  • `m`: number of indexers.
  • `queue_time`: an array of increasing timestamps when documents are queued.
  • `processing_time`: an array containing the time it takes to process each document.

  The algorithm was required to compute:

  • The total number of documents processed successfully.
  • The top `k` indexers that processed the most number of documents (where `k` is a new input).