Background: CS undergrad, competitive programming background (LeetCode Knight)

Sharing my Amazon SDE Intern interview experience, 2 rounds, ~1 hour each, both technical + behavioral.

Round 1

DSA: Daily Temperatures (https://leetcode.com/problems/daily-temperatures/). Brute force O(n²) first, then a monotonic decreasing stack. Wrote it by hand on paper and tested it against his example.

vector<int> dailyTemperatures(vector<int>& temperatures) {
    int n = temperatures.size();
    vector<int> answer(n, 0);
    stack<int> st; // indices still waiting for a warmer day

    for (int i = 0; i < n; i++) {
        while (!st.empty() && temperatures[st.top()] < temperatures[i]) {
            int j = st.top();
            st.pop();
            answer[j] = i - j;
        }
        st.push(i);
    }

    return answer;
}

O(n) time and space. Follow-up: Next Greater Element II (https://leetcode.com/problems/next-greater-element-ii/), the circular array version, know the 2n + modulo trick. He also asked why a stack works here, not just that it does, be ready to explain that.

Behavioral: toughest feedback I'd received and what changed after. Also asked why Amazon specifically (my answer: large-scale data, real production systems). Both interviewers were clearly matching my answers to Amazon's Leadership Principles, so learn all 16 and know which one fits each story you tell.

Round 2

Mostly a discussion of my research work, then GenAI questions and a problem-solving round.

GenAI: what causes hallucination and how to fix it (RAG, step-by-step prompting), why guardrails matter, bias in training data, and using RAG when a model doesn't know about a specific country's context. My example: a model trained mostly on US law getting an Indian law question wrong.

Problem-solving: Reorganize String (https://leetcode.com/problems/reorganize-string/), just talked through it, no code. First check if it's even possible (no character can appear more than (n+1)/2 times), then use a frequency map plus a max-heap, O(n log k). He kept changing the test case to see how I'd react, not just to check if I was right.

Result: Offer.

Takeaways:

• Explaining a data structure choice out loud is a different skill than coding it. Practice that separately.
• Leadership Principles aren't a checkbox. Know which one fits each story.
• Know the pros and cons behind your resume projects, not just the results.
• For GenAI questions, have one real example ready instead of a textbook definition.

Full round-by-round writeup: https://k3tikvats.substack.com/p/amazon-sde-internship-interview-experience

Happy to answer questions, do upvote and like my blog if found useful ;)