A few weeks ago, I was asked to use AI during both a live coding interview and a take-home assessment. At first, I wasn’t sure what to expect because it was very different from the traditional interview process.
In this post, I want to share my experience and a few things I learned along the way. AI can definitely help you work faster. However, interviewers still want to see how you think, make decisions and solve problems. The goal is to use AI to support your work while still demonstrating your own skills.
So what was my experience like?
Because AI was allowed during the interview, the time available actually felt much shorter than I expected. Instead of immediately asking AI to solve the problem, I found it much more valuable to spend a few minutes understanding the task first.
I took the time to read the requirements carefully. Plus, I got the application running and familiarised myself with the codebase. I also explored the user flow, so I understood how everything worked. If the task was to fix a bug, I first reproduced the issue and made sure I understood the expected behaviour before asking AI for help.
I quickly realised that the quality of the AI’s response depended on the quality of the context I provided. The more information I gave, such as the files involved in the issue I had found and the expected outcome, the more accurate and useful the suggestions became.
Understanding the problem before prompting AI
One of the biggest lessons I learned was not to blindly accept the AI’s suggestions. I reviewed every change and made sure I understood what had been modified, why it was necessary and how it solved the problem.
The interview process allowed candidates to use AI during the coding challenge, but it was also clear that understanding every change was still expected. However, I would still need to walk through my solution afterwards. I had to explain my approach, justify my decisions and demonstrate that I fully understood the implementation. AI helped me move faster, but my reasoning, technical understanding and decision-making were still what the interviewers were evaluating.
AI still needs your technical judgement
Another thing I noticed was that AI removed a lot of the mental load around syntax and remembering framework methods. That allowed me to focus more on solving the problem itself.
Even then, I still questioned many of the AI’s suggestions. Sometimes it generated a solution that worked but overlooked important details such as handling undefined values, using more modern language features like the null coalescing operator or early returns where appropriate, considering performance or thinking about edge cases. AI is very good at finding a solution but not always the best solution.
That is where your experience as a developer still matters. You need to recognise these issues, challenge the AI’s suggestions and ask follow-up questions until the solution reaches the standard you expect. I found that when I gave AI better direction, it usually improved the solution within a few minutes.
In the end, AI helped me complete the coding challenge more efficiently. However, I still had to guide it. My responsibility was not just to produce working code. It was to make sure the solution was maintainable, performed well and was easy to understand. Those were the skills the interview was really evaluating.
Presenting my solution
Another important part of the process was presenting my solution. Writing the code was only half of the challenge. The other half was communicating my thought process clearly.
Before the session, I reviewed each task and walked through my solution so I could confidently explain what I had done. I also found it helpful to have a clear story about my approach. For example, I explained that I chose to fix the existing bugs before implementing the new features because, based on my experience, it is usually better to resolve known issues before building on top of them.
While preparing, I also identified a few areas that I would improve if I had more time. Sharing those ideas showed that I had thought beyond simply completing the task and had considered how the solution could be improved further.
One technique that really helped me was leaving a few comments in the code as reminders for myself. These were not comments explaining what the code did. They were simply cues that helped me remember the key decisions I wanted to talk about during the presentation.
I also reminded myself to stay as calm as possible; I will admit this is something I am still working on. I kept telling myself that the goal was not to give a perfect presentation. The goal was to clearly communicate my thinking, explain my decisions and show how I approached the problem.
The live coding session
During the presentation, I was also asked to complete a small coding challenge by adding another feature to the take-home assessment. Once again, I was allowed to use AI and I only had ten minutes to complete it.
To be honest, using AI during the interview almost felt like I was cheating at first. That feeling disappeared once I realised the interviewer was not evaluating whether AI could write the code. They wanted to understand how I used AI as part of my workflow.
Whenever I had doubts about the requirements, I asked questions instead of making assumptions. I wanted to make sure I understood the task before asking AI for help. Additionally, I also tested my solution before saying I was finished. I think this demonstrated that I was not simply copying AI’s suggestions but taking ownership of the solution.
One of the questions the interviewer asked me afterwards was about my experience using AI. My answer was simple. AI is an excellent tool that helps developers move faster. However, it is still your responsibility to understand the code, make good technical decisions and ensure the solution is maintainable, scalable and of high quality.
Final thoughts
The entire experience was definitely nerve-wracking. However, it also changed the way I think about AI in software development.
AI can make you a faster developer. It can help you explore ideas, solve problems and remove some of the repetitive work. What it cannot replace is your ability to think critically, understand the code, communicate your decisions and recognise when a solution needs improvement. AI helped me write the code. I still had to engineer the solution. That’s your individuality.
I believe interviews like this will become more common as AI becomes a bigger part of our everyday workflow. Learning how to work alongside AI without becoming dependent on it is a skill worth developing.
I was really happy to receive an offer after the interview and I left with a much better understanding of what companies are looking for when they allow AI in the interview process. 😄
Thank you for reading. I hope sharing my experience helps you feel more confident in your next coding interview, especially if AI is part of the process.
Good luck and all the best! 😊
