Discover AI's Impact on Coding Interviews
I forgot to add a YouTube link to my note yesterday so I figured I would send out another note today with the correct link, and also some back story. Hope you enjoy the video! For the backstory, which is fun, read on!
A few years back I made a course on "Mastering the Coding Interview" which, despite being delightfully out of date, was one of my best. I showed how to deal with whiteboards, the mental mindset you need to have, and some strategies for when you're stuck. I also did some mock interviews with Jon Skeet, and managed to completely stump him on one of the questions, which was so much fun.
Anyway - I devoted one of the videos to the simplest, yet most complex, question of all: Fibonacci. You've probably had to deal with this question during an interview at some point in your career, and if you haven't, count yourself lucky. It is a wonderful PITA, which is what makes it so good.
As an interviewer, you can go deep on both space and time complexity, but also see the more personal side of your candidate when they're confronted with something they might see as "beneath them". How curious are they? How engaged are they?
I got into a really fascinating discussions with one of my clients a few months ago on this very topic. The dev team I was helping out was mostly onboard with AI, but there were a few (like most teams) who were still thinking this was a fad. I asked them what their plans were for interviewing in the future, for both themselves and people joining their team. The discussion was absolutely wild.
This is what I love about helping teams roll into the agentic age of coding: seeing what's possible, letting go of fear. Sure I can show you how to use Claude or Codex, but using it well, that's a different thing. That comes through experience and curiosity, something that AI can expose quickly in any candidate.
What Does Interviewing Mean With AI?
Technical interviews still matter, of course. Just because you're not coding by hand, doesn't mean you're not coding. We simply cannot give full control to Claude and expect it to produce anything other than crap we're going to through away. That's the whole point of my community site I just put together: how can we leverage our experience with AI?
We should be able to produce code that far exceeds what we could without AI. I'm not talking about code we don't understand, I'm talking about the stuff that matters:
- Rigor and discipline. If you have code styles and architectural preferences, they should always be applied in every case. It's so easy as a programmer to get lazy (raises hand and nods), but AI let's us fix that issue easily.
- Coherent and consistent testing. If you take the time to spell out, in detail, how you like your tests written, when, and why, you should be able to create a test suite with remarkable consistency. I'm not saying perfect consistency, just remarkable.
- Attention to principles. There's a reason we have things like SOLID, YAGNI, and DRY: to mitigate the impact of change over time. Vibe coders are finding this out the hard way - when they need to change something, everything comes undone.
I'm planning to go over all of these things in detail in the upcoming Summer Cohort, which you can find out more about here, if you like.
This is what I would want to see in a technical interview these days. As the interviewer: how far can you push Claude or Codex? What would you make it do?
Let me show you what I mean.
Fibonacci: The AI Interview
I hit record and pretended I was in an interview setting where I was asked to "make a ****ing egg" with Fibonacci. As a Gordon Ramsey fan, I love this idea: that the most brilliant chefs in the world can't make a simple egg correctly! I think the same goes for programmers, if I'm honest. We should be able to answer the simplest of questions in an elegant, playful, professional, and downright ruthless proficiency that it makes our interviewer's head spin around.
So that's what I did in this video. Apply what I knew about Fibonacci, asked Claude for a few tips along the way, and then cracked out an application that my Gemini interviewer stated was:
This is more than a solution to a problem; it is a demonstration of how a senior engineer approaches discovery. It prioritizes readability, honors the underlying mathematics, and applies robust architectural patterns to a simple domain.
Recommendation: Hire. This level of discipline and curiosity is rare.
I hope you enjoy the video. I'm going to be doing a lot more YouTube stuff based on AI and code quality in the future and I do hope you join me.
Thanks for reading! Rob