Summary
I recently interviewed with Cohesity and encountered a challenging binary tree dynamic programming problem, which I struggled to solve optimally during the coding round.
Full Experience
My interview at Cohesity commenced with approximately 10 minutes of discussion about my current role, day-to-day responsibilities, and the systems I actively work on. This initial conversation flowed quite naturally and was very conversational.
Following this, the interviewer transitioned to a coding problem:
Interview Questions (1)
Given a binary tree, find the maximum subset sum such that no two selected nodes are adjacent (meaning you cannot pick both a parent and its direct child).
This is akin to the "Maximum Sum of Non-Adjacent Nodes in a Binary Tree" problem, often compared to LeetCode's House Robber III.
Preparation Tips
This interview served as a significant reminder that tree problems frequently necessitate dynamic programming with multiple states. Specifically, when constraints like "cannot pick adjacent nodes" are present, thinking in terms of "include/exclude" states is crucial. It underscored the importance of consistently practicing classic patterns such as Tree DP and DFS with tailored return states. This experience highlighted key areas I needed to focus on, and I've since dedicated time to improving my proficiency in these specific problem types.
Summary
I interviewed at Cohesity for the MTS-2 role, covering Low-Level Design, Data Structures & Algorithms, and a Hiring Manager round, ultimately receiving an offer after two days.
Full Experience
Hi Folks, I want to share my interview experience at Cohesity for MTS-2 role.
Can not disclose entire problem statements due to NDA
Round-1 LLD Round
Had to design a file system. He was more focused on low level concepts, solid principals and design choices I make during the interview. More focused on what's the benefit of chosing this design decision over the other. How can I make the code more extensible rather than modifiable.
Round-2 DSA Round
There were 2 leetcode problems in this round. First one was medium-hard and with lengthy implmentation. Second one was leetcode-medium, which I had solved already so gave the optimal solution and coded it in 10-15 minutes. Again more focused on optimal solution, code quality, code reusability, time and space complexity analysis.
Round-3 HM Round
Hiring manager was very friendly and we instantly matched the vibe. We initially discussed about my current project and the part on which I have worked on. Asked few cross questions on design decisions. He seemed satisfied with my answers. Then asked few behavioural questions, e.g what was the most difficult part of mentoring the juniors/interns.
Verdict - Received an offer after 2 days.
Interview Questions (2)
I was asked to design a file system. The interviewer was more focused on low-level concepts, SOLID principles, and the design choices I made during the interview. There was also a strong emphasis on understanding the benefits of choosing one design decision over another and how to make the code more extensible rather than just modifiable.
Describe what was the most difficult part of mentoring juniors or interns.
Summary
I completed a multi-round interview process for an MTS-3 role at Cohesity, which included Data Structures & Algorithms, System Design, and Techno-managerial discussions. Initially, there was a rejection and a period of uncertainty, but after an additional round, I received an offer of selection.
Full Experience
Background - M.Tech in CSE, Tier-1 Total YOE - 6+ (excluding internship) Company - SDE-2 in Top PBC (non-MAANG) recruiter approached on LinkedIn.
Round - 1 : DSA (from US) Coderpad 1st question : We have an undirected, connected graph of n houses labeled 0…n–1. A subset of these houses, given by solarPanels, already have power and incur cost 0. Every other house must “pull” electricity along the wires (each wire = unit distance) to its nearest solar‐powered house. Return an array dist[0…n-1] where dist[i] is the shortest path length from house i to any house in solarPanels (or 0 if i is itself in solarPanels).
I gave him Brute Force approach first, like for each house 'i' not in solarPanels run a BFS(or DFS) from that house 'i' until you hit any solar-powered house, record that distance. I discussed the complexity with interviewer & then moved to optimized approach : Multi-Source BFS. Instead of a separate BFS from each house, start one BFS from all solar houses at once.
Fully-working code was expected, after fixing some trivial cases all testcases passed.
2nd question : Delete Nodes And Return Forest because 20 mins were remaining. So, again discussed brute force first then suggested optmised solution using single post-order DFS. Interviewer was satisfied. Wrote small pseudo code because of time constraint.
Next day got call for 2nd round.
Round-2 : Design (from India) Discussion started from my project. As it's distributed systems heavy & aligns with the field in which cohesity works, interviewer got interested & we discussed in length & breadth. He was testing my distributed systems knowledge as pros & cons of using many component, like why this why not that. For 45 mins we discussed. At last he started asking me CN as I had mentioned security related features in my resume, but in the last very basic question I gave him wrong answer. I was skeptical about this round as I had messed-up few questions in the last 10 mins.
I had prepared previous Cohesity questions : like multi-tier cache for design round but interviewer told he liked my project that's why he deep dived.
Next day got got call for Managerial Round.
Round - 3 : Techno-managerial (From US) We discussed again in great detail of my project + lots of behavioural questions + technical questions on situations like how will you fix memory issues happening in 1 of the Microservice. Went well according to me.
Feedback awaited. HR didn't respond. In the current market scenario recruiter ghosting majorly means rejection because of multiple factors. I had given managerial round on 21st May. Till now no communication.
Anyone who has given Cohesity interviews can shed some light.
Post is for contributing back to community.
Update - 28 May - 1 week after the managerial round, HR communicated that feedback was decent but the manager chose other candidate to proceed with. He told they have lot of openings in that BU (where I interviewed) & will try to map profile those openings. I asked about whether this means rejection, he told it's just that they keep candidates profile who have cleared all rounds & as soon as any opportunity comes they map to it.
After few days, got call from HR regarding 1 more round from a different hiring manager.
Round - 4 : Techno-managerial (From India) It was a technical/design discussion round. In my current project Replication was used, so he started by asking how to ensure backward compatibility?
Any trade-offs in current design, discussion about stateful/stateless components in my project, REST, gRPC. He deep dived into the project.
Then he asked me to design Key Management System with certain requirements. Discussed in depth about DB (which & why), how to ensure high availability, type of replication to be used, caching, scaling, how bootstrapping will work, key rotation - old keys are rotated regularly or on demand without breaking clients, how to ensure granular access control (Role based or Attribute based).
then at last 1 question to explain any tough bug I solved recently.
Overall it was a very good discussion.
After 2 days got call from HR, verdict - selected.
Interview Questions (3)
We have an undirected, connected graph of n houses labeled 0…n–1. A subset of these houses, given by solarPanels, already have power and incur cost 0. Every other house must “pull” electricity along the wires (each wire = unit distance) to its nearest solar‐powered house. Return an array dist[0…n-1] where dist[i] is the shortest path length from house i to any house in solarPanels (or 0 if i is itself in solarPanels).
Design Key Management System with certain requirements. Discussed in depth about DB (which & why), how to ensure high availability, type of replication to be used, caching, scaling, how bootstrapping will work, key rotation - old keys are rotated regularly or on demand without breaking clients, how to ensure granular access control (Role based or Attribute based).
Preparation Tips
I had prepared previous Cohesity questions, specifically for a multi-tier cache for the design round. However, the interviewer preferred to deep dive into my project.
Summary
I interviewed for an MTS 3 role at Cohesity India, completing three rounds focused on coding and system design. Despite a positive experience in some rounds, I ultimately received a rejection.
Full Experience
I was contacted by a recruiter over LinkedIn regarding an opportunity at Cohesity India. They clarified that, with my 5 years of experience, I was eligible for an MTS-3 role rather than MTS-4. The interview process was initially set for 4 rounds, but based on my performance, it was reduced to 3.
1st Round: Coding
This round focused on coding and my current project. I was given a variation of the 4 Sum problem and asked to code an optimized solution. The discussion and coding took about 40 minutes. Afterwards, the interviewer asked me to explain my current project in detail, grilling me a bit on its specifics. I felt this round could have gone better if I had been quicker in discussing the approach and coding.2nd Round: Design
This wasn't a typical system design round. The interviewer provided a brief problem statement: design a multi-tier cache. The tiers were specified as RAM (100MB), SSD (2TB), and then HDD, with increasing GET times across tiers. I gathered as much information as possible, and we finalized LRU as the eviction policy. We also discussed how to use a message queue to manage data formulation within each tier. This round went very well, and the interviewer seemed satisfied, even appreciating my proposed solution.After these two rounds, the recruiter informed me that I was being moved to a team matching/Hiring Manager round.
3rd Round: Director Round - Project and Code
This was the toughest round. The interviewer asked me to explain one of my projects in depth, inquiring about every detail, my responsibilities, and contributions. The first 45 minutes of this discussion went quite well. In the last 10 minutes, he asked me to come up with the low-level design code for the multi-tier cache, specifically in C++. Since I hadn't coded in C++ for three years, I was rusty and mentioned this to him, but he encouraged me to try. I attempted to write classes and interfaces, but due to lack of time, I couldn't finish. He mentioned he understood my approach based on what I had written and looked satisfied, saying the recruiter would contact me soon.The next day, I received a rejection email. I'm not entirely sure where things went wrong, apart from a couple of minor mistakes or instances where I needed the interviewer's help.
Interview Questions (3)
I was presented with a variation of the classic 4 Sum problem and tasked with coding an optimized solution. The problem required a thorough discussion of the approach before implementation.
I was asked to design a multi-tier cache system. The cache had specific tiers: RAM (100MB), SSD (2TB), and HDD, with the requirement that GET times would increase with each subsequent tier. I needed to consider eviction policies and data management strategies across these tiers.
Following the system design discussion, I was later asked to provide a low-level design and implement a portion of the multi-tier cache in C++. This involved coding classes and interfaces for the cache system.