FamPay | Tech Lead | Interview Experience | Rejected

I had 7+ years of experience and was working at a startup at the time.

The process consisted of an initial HR conversation followed by DSA and LLD technical discussions.

Round 1: HR Discussion

The first call was primarily an introductory conversation.

We discussed my professional background, previous projects, current responsibilities, and what I was looking for in my next opportunity.

The conversation was straightforward, and I moved ahead to the technical rounds.

Round 2: DSA

The next round focused on problem solving and data structures.

I was given a Longest Consecutive Sequence type problem and was asked to come up with an efficient solution rather than simply implementing the brute-force approach.

We discussed different ways to approach the problem, including the time and space complexity of each solution. I was able to work through the problem and explain the reasoning behind the final approach.

The interviewer also asked a few follow-up questions around edge cases and how the solution would behave with larger inputs.

This round went reasonably well, and I moved on to the LLD discussion.

Round 3: LLD - Rate Limiter

The next technical round started with a short introduction before moving into the main design problem.

I was asked to design a rate limiter that could apply different request limits to individual users.

Before starting the implementation, we discussed various rate-limiting algorithms and the situations where each one would be appropriate.

I decided to use the Sliding Window Counter approach.

The basic idea was to maintain request counts for the current and previous windows and calculate the effective request count using:

current_count + (previous_window_percentage × previous_count)

Where I Struggled

Although I understood the approach conceptually, I had difficulty translating the core algorithm into working code during the interview.

I haven't been writing this kind of low-level implementation manually as often since using AI coding tools such as Claude more regularly in my day-to-day work.

I was able to implement the user-specific threshold logic and several other parts of the rate limiter, but the central sliding-window calculation wasn't working correctly.

I spent the remaining time trying to fix the implementation, but I couldn't get it completely right before the round ended.

Looking back, I should probably have chosen the Sliding Log technique. It would have been somewhat easier for me to implement correctly under interview pressure, even though it has different memory and performance characteristics.

The interviewer asked if I had any questions at the end. By that point, I had a fairly good idea that the round hadn't gone well, so I didn't have any questions and wrapped up the call.

Final Result

I received the rejection the following day.

The main lesson I took away from this interview was that senior-level interviews test more than conceptual knowledge. Being able to implement the underlying logic yourself, reason through trade-offs, and choose an approach that you can confidently execute under time constraints is equally important.

For me, it was also a reminder that while AI tools can make day-to-day development much faster, it is still important to regularly practice writing core algorithms and logic from scratch.