Data Structures & Algorithms - Preparing for Interviews

Overcoming Interview Anxiety

I won’t call it a fear, but I would like to overcome interview anxiety. I do not enjoy DSA interviews, it comes with some performance anxiety for me. I overcame my fear of swimming, and that’s a real fear for me. I get anxiety in interviews because I don’t enjoy them, and I decided to address this.

My plan is gradual progress. Leetcode is a lot of pattern recognition, and it is very targetted towards knowing what tools you have at your disposal before you attempt the problems. Just staring a Leetcode problem in the face for a few hours will not help you if you do not know you can code a specific way to solve this. I want to code out different data structures and algorithms step by step. To that aim, I’m maintain this repo.

I’ve tried this before, and I’ve never really enjoyed preparing for interviews. However, this time, I have purpose. I am trying to steer my career towards systems and performance engineering and I do not have the luxury of denying DSA interviews anymore. I do not want to take any job I get any longer, it has to meet my standards, and when it does, I will have to be ready to ace any manner of interview that the company desires to throw at me.

To that end, my approach is simple, gradual progress. I’ve asked Claude to come up with a list of exercises for me, small ones. Starting with stuff as small as “print every item of a list”, “print the largest item in a list”, “find if an item exists in a list”, “sum a list”, repeat for linked lists, maps, trees, graphs and so on. Then, I asked it to use incremental progress to draft up a TODO list of sorts, which is maintained in that repo in the README. I want to do this in Python, Javascript, and Rust. I’m not aiming for 100% completion, of course, besides in Python. In Rust, I do not want to waste time trying to build dynamic data structures, no matter how tempting that may be.

Following My Progress

I originally wanted to track my progress here, but I have a very simple way of doing that now, by linking to the public image, which is what the header image of this post is.

This image will be constantly updated in the repo to showcase what I’m learning and when it was last updated, the story of how I got to this image is pretty interesting, but not relevant to the task of learning. Yet, I’m going to share it below.

How I Visualize Progress using Claude

I was using Claude for housekeeping in that repo:

• Updating test fixtures for my code, to ensure that it has all the edgedcases for my “gradual learning exercises”
• Updating tooling for the repo so I can run tests using a justfile without worrying about what broke and when.
• Generating placeholder functions for me to fill in, ensuring those functinos do not have type hints (in python), but just enough of a docstring to get me started.
• Updating the TODO section of my README so that I can focus on doing the exercises

When doing the last step in the above list, I realized I could now plot the progress in some way to show it off.

I asked Claude to make a single PNG file that had two plots:

1. A progress-bar view that showed different topics and how many exercises I was doing on them.
2. A github-style commit view that showed when I was actively studying and highlighting my streaks.

At the first attempt, it used Pillow to generate the image, and I dind’t bother saving that image since it looked pretty boring. I asked it to do the same thing using Matplotlib, and it made this.

I liked it, it was more than I could sit and do in matplotlib myself (I do not enjoy writing any matplotlib code!), and I was happy making tiny adjustments to it.

I next tried to have it clean up the legends, make the charts more visible and suggested some size/position improvements. At the same time I asked it to clean up the matplotlib code and write it in the eyes of someone who’s seasoned in data viz and in the intracracies of matplotlib.

This is what I got when I asked it to “think like a data storyteller”.

Then, my friend recommended trying the frontend-design skill to ask it to redesign the graphic from the ground up, and the simplify-code plugin to make the code more modular and less convoluted.

I was wowed by this. I’m not anything of a data visualization expert, and I’m sure a human could do a far, far better job, but this was really good for me. If it makes studying DSA easier for me, that’s tokens well spent for me.

If you’d like to look at the code that visualizes this, be warned, it’s completely vibe-coded and plenty esoteric, but it’s available here