Amazon SDE2 Interview Experience
- Posted
- 31 Mar 2026
Summary
I went through the Amazon SDE-2 interview loop, completed several rounds including DSA, system design, and behavioral questions, but the final HLD round was cancelled and I was not considered further.
How it went
Went through the Amazon SDE-2 interview loop recently.
Round 1 was supposed to be HLD, but it got rescheduled due to unavailability of Interviewer and then just never happened.
Round 2 was DSA - Started with LP discussion for 20-30 mins and after that it was a hard DP problem.
Round 3 was more like LLD + general problem solving. Focus was on writing clean, maintainable code. Had LP questions for 20-30 mins and a good discussion on designing a rate limiter and its different algorithms token bucket, fixed window, sliding window along with implementation of one of them.
Round 4 started with ~10 mins of behavioral (LPs), then DSA:
- LCA variant
- LRU cache with TTL Implemented both and discussed time complexity.
Then came the final HLD round which was supposed to happen.
A day before my HLD round, got a call saying it’s cancelled. That was it. No explanation, no feedback. Later saw the portal updated to “no longer under consideration” — no email, nothing. I did my best in all three rounds.
Not gonna lie, it’s a bit disappointing after putting in so much time and effort, and then not getting any feedback at all. But yeah, I’ll take it as a learning experience.
Questions asked (2)
- 01
LRU Cache with TTL
Data Structures & Algorithms
Implement an LRU (Least Recently Used) cache that supports a Time‑To‑Live (TTL) for each entry. The cache should provide
get(key)andput(key, value, ttl)operations, evict least‑recently used items when capacity is reached, and automatically expire entries after their TTL. Discuss the time and space complexity of your solution. - 02
Design a Rate Limiter
System Design
Design a rate‑limiting system and discuss various algorithms such as token bucket, fixed window, and sliding window. Implement one of these algorithms to enforce a limit of N requests per time interval.