He cleared every DSA round at Amazon.
Rejected anyway. Reason: design depth.
He walked me through the whole loop yesterday on a 1:1 call.
Here is how it actually went, in order.
He prepped the right way, for what it is worth. Read every 2026 Amazon writeup he could find. Took notes on patterns instead of grinding random problems.
Asked for every round on a different day so he got a fresh head each time.
Then the schedule landed.
ROUND 1, BAR RAISER:
- Every writeup he had read put the Bar Raiser at the end.
- His was first.
- And it was LLD. The one area he was weakest in.
The question was a real problem their team was working on.
Audio buffers moving across processing stages at a fixed frame rate.
- He did not understand it for the first ten minutes.
- Took hints. Got into a good discussion.
- Drifted into the HLD of the thing.
- Then looked up with ten minutes left and nothing on the screen.
Nothing.
ROUND 2, DSA:
- File system storage cache.
- He spotted it fast. Storage is not the problem. Eviction is.
- LRU. Doubly linked list plus hash map. O(1) get and put.
Mid coding the interviewer asked him:
when you move a node, do you delete and recreate it, or just reposition it?
- He second guessed himself. Said delete and recreate. Then talked himself back.
Then she changed the requirement. Store file content, not paths.
Time ran out before he finished.
ROUND 3, HIRING MANAGER, HLD:
- Spotify wants to use Alexa to notify users about a new feature.
- He built the happy path first, on purpose. Right call.
The interviewer spent the rest of the round on failure cases.
Mostly one: what happens when the device is offline.
ROUND 4, DSA:
Course dependencies.
Prerequisites. At most K courses a semester.
Priority by the deepest chain of work hanging off each course.
Find the minimum number of semesters.
DFS for depth. Max heap to prioritise. Topological sort to process.
He checked alignment three times. She confirmed every time.
Coded it fully. She said she liked his thinking.
Then he went home and worked out that his greedy was probably wrong.
What matters could be the whole descendant set, not just the deepest chain.
Nobody in the room caught it.
He caught it four hours late.
Three days of silence. Then the rejection call.
He called it shattering.
Here is my read.
He lost this loop in the first hour, not the last.
Bar Raiser first. Weakest format. A problem he could not parse for ten minutes.
Everything after that was him climbing out of a hole.
And strong DSA does not fill a design hole.
There is no exchange rate between those two rounds.
If your loop opens with the round you dread, that was the round you had to prepare hardest.
๐๐ฒ๐ฒ๐ฝ๐ถ๐ป๐ด ๐๐ต๐ถ๐ ๐ถ๐ป ๐บ๐ถ๐ป๐ฑ, ๐ ๐๐ฒ๐ป๐ ๐ฑ๐ฒ๐ฒ๐ฝ ๐ฎ๐ป๐ฑ ๐ฑ๐ผ๐ฐ๐๐บ๐ฒ๐ป๐๐ฒ๐ฑ ๐ฒ๐๐ฒ๐ฟ๐๐๐ต๐ถ๐ป๐ด ๐ถ๐ป๐๐ผ ๐ฎ ๐๐ฎ๐๐ฎ ๐๐ฎ๐ฐ๐ธ๐ฒ๐ป๐ฑ ๐๐ฒ๐๐ฒ๐น๐ผ๐ฝ๐ฒ๐ฟ ๐๐๐ถ๐ฑ๐ฒ.
๐๐ฒ๐ ๐๐ต๐ฒ ๐๐๐ถ๐ฑ๐ฒ ๐ต๐ฒ๐ฟ๐ฒ: https://lnkd.in/dTvYVutD
Use SDE30 to get 30% off.
Stay Hungry, Stay FoolisH!
2August 30, 2026 616
