← Back to archive

Using AI in interviews reveals real engineering skill

Usage of AI during an interview.

LLMs aren’t magic. They just reflect how skilled you are. If you’re good, they amplify that. If you’re lost, they don’t save you. It’s kind of wild how quickly things have shifted - someone might’ve been a top-tier engineer in 2024, but if they haven’t adapted to AI in 2025, they’re already behind.

The part that stuck with me most? The idea of watching someone dance with an LLM during an interview. Not just answering questions - but actually seeing how they explore, debug, iterate, and use the model like a tool, not a crutch. That’s the real signal.

  • Don’t ban AI for an interview. That just forces people to hide it.
  • Ask better questions - like which LLM they’d use for XYZ and why.
  • Look for curiosity. Have they tried building agents? Do they understand context limits? Have they hit real-world pain and built something to fix it?

It’s not just about what they say - it’s how they work. Do they explore? Ask smart questions? Reuse prompts? Think in loops?

Originally posted on X

PS: I'm pretty active on 𝕏 if you'd like to follow along

Archive

2026

2025

2024

2018