Nutanix

Role: MTS-1

College: Decent uni (BTech CS)

Interview Date: July 2024

Interview Type: On-site (after OA)

Result: Rejected

Difficulty: OA: Easy-Medium. DSA: Medium. Debugging: Medium. System Design: Open-ended (Medium-Hard)

Rounds: 3 rounds (Online Assessment, DSA, Debugging and System Design)

Topics Asked: Heap, DFS, Tree problems (inverted level order sum), Graph problems (dependency graph, cycle detection), Debugging (linked list palindrome), System Design (scalable data request system, API server, load balancer, database scaling, caching strategies - LRU, LFU, cache-aside, read-through), Resume Discussion

Experience:

The candidate, with 1 year of experience as an SDE, interviewed for the MTS-1 position at Nutanix in Bangalore.

Round 1: Online Assessment (OA): A 2-hour HackerRank challenge with two questions: one easy and one medium, based on heap and DFS. All test cases were passed.

Round 2: DSA: This on-site round involved two questions.

• Q1: Given a tree, multiply the value in each node with its inverted level order and return the sum of all these values. The candidate suggested a DFS approach, storing elements of the same level in an array, then summing them after multiplying with reverse levels. Optimization was requested to directly store the sum of each level. Code and dry-run were performed.
• Q2: Given a dependency graph (e.g., A -> B -> C), print an order of execution such that all dependency items are executed first. If a cycle exists, throw an error. A straightforward DFS approach with cycle detection was implemented. The interviewer seemed satisfied.

Round 3: Debugging and System Design:

• Debugging: The interviewer presented Java/C++ pseudo-code for checking if a linked list is a palindrome. The code had logical errors, which the candidate successfully identified and dry-ran test cases for. This took 20 minutes, and the interviewer seemed satisfied.
• System Design: An open-ended question was posed about designing a system for many users requesting "X" data. The candidate started with a basic API server, load balancer, and database, then scaled components horizontally. Database scaling involved partitions and replication. Further scaling included caching, with discussions on strategies like LRU, LFU for invalidation, cache-aside, and read-through. For database speed-up (relevant to a DB development team role), an integrated cache with the DB query engine was suggested. The round also included a resume walkthrough. The interviewer appeared satisfied, and 10 minutes remained.

Despite feeling that the rounds went well, the candidate later received a rejection call from HR, stating the first round was good but the second round was "so-and-so."