Kotak Mahindra Bank SDE 3 Interview Experience | Hyderabad |

My interview experience at Kotak Mahindra Bank for the SDE 3 role in Hyderabad involved a mix of practical coding and problem-solving.

Bar Raiser - Low-Level Design (LLD)

I was asked to design and implement a Battleship game using Low-Level Design principles.

The expectation was to write complete, executable code in an IDE rather than just explain the design or provide pseudocode. I had to focus on structuring the game, identifying its core components, and implementing the functionality in a way that could run successfully.

The interviewer also provided an IDE, but I found the environment difficult to work with. I would recommend politely asking whether you can use a local IDE if the provided setup is uncomfortable. In my experience, the interviewer was cooperative, so it is worth asking.

This round required me to balance design decisions with actual implementation, as the goal was to deliver working code within the allotted time.

Round 2: Data Structures and Algorithms (DSA)

In this round, I was asked a dynamic programming question about maximizing the sum of selected coins without choosing adjacent coins. The interviewer first presented a one-dimensional problem and then extended it to a two-dimensional grid.

Question 1

I was given an array representing coins arranged in a single row. My task was to select coins such that their total value was maximized, with the restriction that I could not select two adjacent coins.

Example:

Input: coins = [4, 5, 6, 7, 8]

Output: 18

Explanation:

I could select the coins with values 4, 6, and 8, located at indices 0, 2, and 4.

The maximum sum for this example was:

4 + 6 + 8 = 18

The main idea was to decide at each index whether selecting the current coin and skipping the previous one would give a better result than excluding the current coin. I could approach this using dynamic programming by tracking the maximum sum achievable up to each position.

Question 2

The interviewer then extended the problem to a grid containing two rows of coins, each with the same number of columns.

My task was still to maximize the total value of the selected coins, but the adjacency constraints now applied in all eight directions:

• Horizontally: Left and right.

• Vertically: Directly above and below.

• Diagonally: Top-left, top-right, bottom-left, and bottom-right.

Whenever I selected a coin, I could not select any other coin adjacent to it in any of these directions.

This variation was more challenging because selecting a coin in one row could restrict my choices in the other row, including diagonal positions. I had to think beyond the standard one-dimensional recurrence and consider how valid selections across consecutive columns affected one another.

The problem tested my ability to recognize the underlying dynamic programming pattern and adapt it to a two-dimensional setting.

I was ultimately rejected after these rounds. The experience highlighted the importance of practising both complete LLD implementations and dynamic programming problems with follow-up variations.

I am sharing these questions to help others preparing for SDE 3 interviews, particularly for Java backend roles.