Amazon (AWS) | SDE 2 (L5) | India | Reject

✅ Online Assessment
submitted on 30th April

📲 Phone screen
happened on June 05
After 11 days, on June 16, I was informed moving forward to final loop.

3 Rounds on June 23 (F2F)
4th Round on June 24

🔢 Round 1 - DSA Round
Rotten Oranges (Multi source BFS)

It began with some technical questions on past experience. Some of the questions were asked relevant to the technology I used but a bit outside of my scope of work where I mentioned I dont have much knowledge in that area. I gave full functional code for the given problem and mentioned one optimisation in space we can do.

2 LPs

🏗 Round 2 - High-Level Design (HLD)
Design News Feed System

I have given the design and explained the flow very well. Couple of mistakes I did were:

• I apparently haven't focused on storing posts but told that I would generate feed after each post and store in DB. I could have made it simpler by choosing to store posts in DB and then through event, I could have generated feed for each post.
• Second, eventhough I was aware that to store the relations between followers and followees, I should use Graph DB but I dont know why I proposed key-value store. It could be probably because of interview stress. I still feel embarassed myself.

2 LPs

🧱 Round 3 - Logical and Maintainable Code
Design Splitwise

Question was not direct. It was a problem statement - Given N people and (person-id, txn amt), min txns to settle among people.

I honestly didn't do this question before. So I proposed a greedy way of using 2 Max Heaps (debters, creditors) - always fetch highest debter and highest creditor and push back to maxheap, if still due is pending. Interviewer pointed out that it doesnt work for all testcases but still asked me to write pseudo code.

I later figured that was an NP-Hard problem and could've used a brute force - backtracking approach. O(N!)

2 LPs

💡 Round 4 - Bar Raiser
Design API to respond with Latency under 2 seconds where data is present in Amazon S3 Database where user queries on hourly/weekly basis

I was expected to implement a system that satisfies those needs.
I had used streams to do that and store in Time Series DB. Where the API requests will be handled by a microservice who queries the DB.

Initially I had given aggregation responsibility to this microservice but panel reminded about latency when user queries. So I shifted this responsibility to the service that "listens to streams and store in Time series DB". Interviewer was satisfied then.

Bar raraiser was supposed to be LLD as per what recruiter mentioned, but I was asked a HLD.

2 LPs

Experience of LPs:
I believe I answered them well. There were no negative points mentioned in final feedback.

📝 Overall feedback: Rejected.

Recruiter mentioned there were couple of technical gaps identified "that are required for this role" (quotes because recruiter stressed on that)

No detailed feedback shared by recruiter as per norms.