There's a lot of questions you can throw at a developer, from grilling them about their resume to asking how to escape a giant blender. Robert Half, which focuses on IT recruitment, offers a simpler guide to figuring out what you actually need to know, while skipping out on some of the more esoteric paths an interview can take.
1. Please describe the architecture of your most recent project.
This is a great question for parsing how an employee thinks: Did they actually plan out their last project? Can they paint you a picture at a high-level, and did they take into account all the potential angle surrounding the project's scope? One thing to also look out for is their communication skills: Someone who can more effectively convey their work, and better understand the needs of others, can be much more effective in driving progress than a top coder who misunderstands the assignment and plows forward in unproductive directions.
2. What lessons have you learned from your current project?
A few CS courses or a coding bootcamp should not be the end of a good candidates education; ask this question to see how they think about their development. Are they learning from the challenges they've had to overcome in the past tackling large projects, or are they making the same mistakes over and over again?
3. See some code
In 2007, a dark truth came to light: The vast majority of developer candidates can't even program. There's a reason that, for developers, GitHub is quickly becoming the new LinkedIn (for better and worse: Just having a Github repo with one cloned project is liable to get you a few pieces of recruiter spam a week). Be sure to ask for some example code that they can walk you through, and have at least some live coding exercises — though be sure to make this as comfortable for the applicant as possible. Live coding exercises, even simple ones, can be high stress, and there's been a number of stories of top-notch engineers "weeded out" due to test anxiety.
What do you think? Is three interview questions enough to separate the best from the rest, or are there other questions you think provide unique insights?