Razorpay

Role: Senior Software Engineer

College: Not Specified

Interview Date: April 2025

Interview Type: Off-Campus

Result: Rejected

Difficulty: Hard

Rounds: 2 rounds

Topics Asked: Machine Coding (in-memory search engine), HLD (LeaderBoard Design, database choice, caching, security, inverted index, suffix/prefix tree)

Experience:

The candidate, with 2 years and 10 months of experience, applied for a Senior Software Engineer role via referral. The process involved two main rounds.

Round 1: Machine Coding (90 Minutes)
This round included an introduction, question explanation, running test cases, and explaining the solution. The question involved designing and implementing an in-memory search engine to support search functionality on blog content. Requirements included creating datasets/categories, inserting and deleting documents (text), searching for patterns in a given dataset, and ordering search results. This is noted as a frequently asked question at Razorpay. The candidate implemented a simple CRUD application and was able to run all 4 requirements. The interviewer then asked about inverted index, which the candidate explained, and suggested solving this problem via Inverted Index and having knowledge of suffix & prefix trees for preparation.

Round 2: HLD Round
The question was to design LeetCode, with a focus on LeaderBoard Design. The candidate presented a basic CRUD design along with some security and caching optimizations. The interviewer had various follow-up questions, but they ran out of time, and the candidate couldn't answer all of them. The candidate acknowledged that their design was "not the best" and noted a learning point: to do basic CRUD design faster in HLD to allow more time for system optimization and answering interviewer questions. Regarding database choice, the candidate stated that both SQL and NoSQL would work, which they felt was a fine answer but suggested having specific pros for the chosen DB.

The candidate was rejected after this round and shared their HLD diagram, noting much of the explanation was verbal.