Amazon SDE II Interview Experience + Need Honest Feedback
Summary
I completed Amazon's interview process for an SDE II role, progressing through an online assessment, several DSA rounds, a low‑level design, and a high‑level system design interview.
Full Experience
Hi everyone,
I wanted to share my recent Amazon interview experience and get some honest feedback from the community on my chances moving forward.
How it started
I initially took referrals for 5–6 roles, but all of them got canceled. But surprisingly, I later received an OA link from a team I hadn’t even applied to.
So yeah… a bit unexpected, but I went ahead with it.
Online Assessment (OA)
- Q1: Variant of inversion count → I used Fenwick Tree (BIT)
- Q2: Geometry-based problem
The OA went well overall. I’m not sure if there was a simpler approach for Q1, but Fenwick Tree worked.
First Attempt – DSA Round (Elimination Round)
This round was clearly mentioned as eliminatory.
- Reverse Linked List in K groups
- Insert Delete GetRandom O(1) (no duplicates)
I solved both questions, explained approaches, did dry runs, and the interviewer seemed aligned. Leadership principles also went well.
But here comes the twist…
After 2-3 days of no updates, I reached out to the recruiter and was told:
- The position got filled
- My interview feedback was not submitted / lost
At that point, I genuinely felt a bit deserted, because the round had gone well, but it didn’t even count in the process.
Thankfully, the recruiter said she would try to match me with another team, and I agreed to go through the process again.
Second Attempt – New Team
This time, I had 2 rounds scheduled on the same day (DSA + LLD).
Round 1 – DSA
- Trim a binary tree to make it a complete binary tree + maintain a trash queue → Solved using BFS
- Number of islands
Both were solved cleanly with dry runs and edge cases. LPs went well too.
Round 2 – LLD
Design WhatsApp Read Receipts (ticks)
Initially, I approached with design patterns like:
- Observer
- State
- Prototype
But the interviewer clarified:
“Assume WhatsApp is already built, (observer pattern is already built), just design the feature.”
Then moved to HLD concepts:
- Data flow of messages
- Redis Pub/Sub vs Kafka discussion
- Message storage
- Handling offline users
Honestly:
- LLD part felt average
- HLD discussion was strong
Recruiter called me next day and told me that I was qualified for the next rounds.
Next Round – HLD (with SDM)
Design FastTag system
Since I wasn’t very familiar with it, I:
- Started with requirements clarification
- Defined APIs, entities
- Designed system step-by-step
Discussed:
- Idempotency in payments
- Redis for caching most frequent vehicle details
- Scalability
This round went well, and the interviewer seemed convinced.
Final Round (Technical – Possibly Bar Raiser)
This was the final round, and I suspected this could be the bar raiser, so I really wanted to perform strongly here.
Coding:
- Longest Substring Without Repeating Characters
- Explained brute force
- Implemented sliding window (optimal)
- Serialize & Deserialize Binary Tree
- Implemented using BFS approach
Where I’m concerned is that the time was very tight—I was able to complete the code, but I didn’t get the chance to do a proper dry run, validate edge cases, or even ask questions at the end.
The interviewer did mention that a DFS approach might have been faster to implement. I explained my reasoning for choosing BFS, but overall, the round felt just okay-ish, not as strong as I would have liked—especially considering this could have been the bar raiser round, which I was hoping to use to compensate for my LLD round that didn’t go as well as expected.
Given this, what do you think are my chances of getting selected?
Interview Questions (8)
Reverse Linked List in K Groups
Given a singly‑linked list, reverse the nodes of the list k at a time and return its modified version. If the number of nodes is not a multiple of k, leave the remainder as is. I solved it by iteratively reversing each group of k nodes.
Insert Delete GetRandom O(1) (No Duplicates)
Design a data structure that supports insert(val), delete(val), and getRandom() in average O(1) time, assuming no duplicate values are inserted. I explained the approach using a hash map to store value‑to‑index mapping and an array (vector) to store the values.
Trim Binary Tree to Complete Binary Tree + Maintain Trash Queue
Given a binary tree, trim it so that it becomes a complete binary tree while also maintaining a "trash" queue for nodes that are removed. I solved it using a breadth‑first search (BFS) traversal to identify missing positions and reorganize the tree accordingly.
Number of Islands
Given a 2‑D grid of '1's (land) and '0's (water), count the number of distinct islands. I used a standard DFS/BFS flood‑fill algorithm to explore each island and increment the count.
Longest Substring Without Repeating Characters
Given a string, find the length of the longest substring without repeating characters. I first explained a brute‑force solution and then implemented the optimal sliding‑window approach.
Serialize & Deserialize Binary Tree
Implement functions to serialize a binary tree to a string and deserialize that string back to the original tree structure. I chose a BFS (level‑order) approach for both operations.
Design WhatsApp Read Receipts (Ticks)
Design the feature that shows read receipts (single tick, double tick) in a messaging app assuming the rest of WhatsApp is already built. Discussed observer patterns, state handling, data flow, and choice of messaging infrastructure (Redis Pub/Sub vs Kafka).
Design FastTag System
Design a high‑level architecture for an electronic toll collection system (FastTag). Covered requirements gathering, API design, entity modeling, idempotent payment processing, caching vehicle details with Redis, and scalability considerations.