Walmart Interview Experiences

1. 💻

   Round 1 — DSA

   A recruiter messaged me on a Friday and the first round was set up for the following Tuesday morning, a one-hour video call. After a short intro the interviewer pasted the problem: return the nodes of a binary tree level by level. I said I would use BFS with a queue and explained the trick of reading the queue size at the start of each level so I knew how many nodes belonged to it. Then I wrote the code, ran it on the sample tree in my head and got [[3],[9,20],[15,7]]. That was the easy part. He then spent almost twenty minutes on complexity — why O(n) time, why the queue can hold up to n/2 nodes at the widest level, and whether a recursive DFS with a depth index would use less memory. I said DFS would use stack space equal to the height instead, so it depends on the tree shape. He seemed happy with that and we closed with two questions about my current project.

2. 💻

   Round 2 — DSA

   Round two was the next day, same format, a different interviewer. She gave me an undirected graph as an adjacency list and a start node and asked whether every node was reachable from the start. I wrote a recursive DFS with a visited set and compared the visited count with the total node count. She asked me to walk through it on her example, and I did, node by node. Then she said, no more code, just talk: how would you count the number of connected components? I explained that I would loop over all nodes, and every time I found an unvisited one, start a fresh DFS and add one to the count. She asked why DFS rather than BFS, and I said both work here, DFS was just less code, and that BFS would matter if we needed shortest paths. The last ten minutes were about recursion depth on a very large graph and switching to an explicit stack. It ended on time.

3. 🧩

   Round 3 — LLD

   The LLD round was two days later and ran about seventy minutes. The prompt was a restaurant system that handles table reservations and also the order flow from placing to serving. I started by asking what a valid reservation was and how walk-ins should be treated, then listed the entities: Customer, Table, Reservation, Order and Waiter. I kept booking and serving as two separate modules and drew the classes in a shared doc. For table allocation I used a strategy interface, and for the order lifecycle a state pattern, and he asked me to explain why for each instead of just naming them. Then he asked how I would actually build it in Spring Boot — services, repositories, how the strategies would be injected — and how a new order type would be added without touching the reservation code. The last part was a coding exercise: print numbers in sequence using three threads. I did it with a shared counter, a lock and a condition, and we discussed what would happen without the lock.

4. 🏗️

   Round 4 — HLD

   The final technical round was the next morning, one hour, with a senior engineer. The problem was to design a video-streaming platform like Netflix or Hotstar. No numbers were given, so I spent the first ten minutes on requirements and then made my own estimates for catalog size, storage per title and peak concurrent viewers, and said out loud how I got each number. Then I drew the main flow: uploads going into object storage, a transcoding step that produces multiple bitrates, a catalog database for metadata with a cache in front, and a CDN serving the actual video segments. She pushed hard on trade-offs — what happens to cost if I keep every quality level in hot storage, what happens to latency if I move things to a cold tier, and where I could accept eventual consistency. I kept saying which option I would pick and why rather than listing both. The recruiter called with the offer within a few days, so the whole process took about a week.

Key takeaway: Interviewers weighted the 'why' behind an approach as heavily as the working code — being able to defend design decisions and reason through follow-ups verbally mattered a lot. The whole process completed within about a week.

1. 💻

   Round 1 — DSA

   I applied through a referral in late February and the first round came about ten days later, a one-hour video call with two problems. The first was group anagrams: given a list of words, put the ones that are anagrams of each other together. I explained that I would sort the letters of each word and use that as a hashmap key, wrote it in Java in a few minutes, and mentioned that a character-count key would avoid the sort. The interviewer asked for the time complexity and I gave both versions. The second problem was the longest substring without repeating characters. I started with the brute force, said it was quadratic, then wrote the sliding-window version with a map from character to its last index and moved the left pointer forward when I saw a repeat. He gave me a couple of edge cases, an empty string and a string of one repeated letter, and both worked. I finished with about five minutes to spare and he said the next round would be LLD.

2. 🔒 2 more rounds — continue this journey in the full kit

1. 💻

   Round 1 — DSA

   I already had an offer from another company when a recruiter reached out, so I told her that up front and she still scheduled the first round for that week. It was a 45-minute call with one problem: find the longest substring that has at most k vowels. I first clarified whether uppercase letters counted and whether k could be zero. Then I said this looked like a sliding window with a counter, and wrote it: move the right pointer, add one to the count if the character is a vowel, and while the count is above k move the left pointer and subtract. I tracked the best length as I went. The interviewer asked me to dry-run it on a short string with k equal to one, and I did that on the shared editor line by line. He then asked what changes if the question becomes exactly k vowels instead of at most k, and I explained it as at-most-k minus at-most-(k−1). We finished a few minutes early and I got the pass mail the same evening.

2. 🔒 2 more rounds — continue this journey in the full kit