A hiring manager shares how software engineering interviews should evolve in the AI era. Rather than LeetCode puzzles, the approach focuses on three stages: a screening conversation that probes fractal knowledge depth about past projects, a system design interview that reveals hands-on vs. theoretical understanding, and a practical coding exercise using a real stubbed codebase where AI tools are explicitly encouraged. The coding interview tests whether candidates can navigate ambiguity, ask business questions, read documentation, and ship working code with AI assistance — not just generate plausible-looking output. The core argument is that AI raises the bar: engineers must own and be accountable for the systems they build, so interviews should test judgment and problem-solving depth, not typing speed or memorized algorithms.
Table of contents
Why you need great engineersInterviews aren't perfectLook for knowledge expanding fractallyGive them all the tools and see how they do3 Comments
Sort: