Amazon SDE 26NG Four-Round VO (9/4, Offer): Three Behavioral Rounds + Path Sum II

Overview

I finished four rounds over the past couple of days and got an Amazon offer in September 🎉 Sharing the details first.

Only one of the four rounds was technical; the other three were about how you handle problems.

Round 1: Behavioral

Self-introduction, then:

• Share a technical disagreement and how it was resolved.

Follow-ups: How do you do the data migration? What's the rollback plan? Why not switch all traffic at once?

Round 2: Behavioral (no coding)

• Tell me about a feature that launched with a poor user experience.

Follow-ups: How did you locate the bottleneck? How did you measure the improvement? How did you trade off quality against speed?

Round 3: Behavioral (no coding)

• Have you had a problem after a launch?

I talked about an API incident. The interviewer asked how it was resolved; I explained that we first communicated the risk, then split degradation into core and non-core APIs so the core path stayed healthy.

Round 4: Coding — Path Sum II

Problem: given a binary tree and a target sum, find all root-to-leaf paths whose sum equals the target.

Approach: DFS through every node and save results via backtracking. To optimize space, keep a single shared path list and pop on the way back instead of copying the list at each level.

def path_sum(root, target):
    res, path = [], []

    def dfs(node, remain):
        if not node:
            return
        path.append(node.val)
        remain -= node.val
        if not node.left and not node.right and remain == 0:
            res.append(path[:])
        dfs(node.left, remain)
        dfs(node.right, remain)
        path.pop()                      # backtrack

    dfs(root, target)
    return res

Follow-up: what if nodes can be negative? — With negative values you can't prune early when the running sum exceeds the target, because later negatives may bring it back down; you must reach a leaf to decide. The code above doesn't rely on pruning, so it works as is.