If you’re preparing for a technical role, you’ve probably spent hours on LeetCode. But here is the catch: not every company uses the “traditional” algorithmic approach. In my time as an Engineering Manager at Google and beyond, I’ve seen the “coding interview” evolve into five distinct styles.
Knowing which one you’re walking into is the best way to lower your heart rate and sharpen your prep. Let’s break down the landscape.
1. The “Traditional” Algorithmic Interview
This is the classic Cracking the Coding Interview style. You’re given a problem statement and asked to write a function to solve it on a whiteboard or a shared editor.
- What they’re testing: Your grasp of data structures, algorithms, and how you communicate your thought process under pressure.
- Pro-Tip: Don’t just jump into the code. Talk through your brute-force solution first, then optimize if you have time.
2. The “Code Review” Interview
Instead of writing code, you’re given a Pull Request (PR) and asked to review it. You’ll walk the interviewer through your comments and suggestions.
- What they’re testing: Your attention to detail, your knowledge of best practices, and—most importantly—your “soft skills” in how you give feedback to teammates.
3. The “Refactor” Interview
You’re handed a snippet of “bad” code—usually bug-ridden and poorly structured—and asked to make it better. You’ll have to “forensically” determine what the code is supposed to do before you can fix it.
- What they’re testing: Your ability to read other people’s code (a massive part of the actual job) and your sense of “clean code” principles.
4. The “Fill in the Blanks” Interview
In this setup, you’re given a large block of code with specific chunks missing. You have to use the surrounding context to determine what’s missing and complete the logic.
- What they’re testing: Your ability to navigate a larger codebase quickly and understand existing patterns without being given the full manual.
5. The “AI Assisted” Interview
This is a relatively new set up and is evolving quickly. I would expect more companies to start to adopt / offer this type of interview as it aligns with the direction of the industry as a whole. From what I’ve heard so far, candidates are getting these types of interviews at Meta, LinkedIn, and Microsoft.
- What they’re testing: Your engineering judgment when utilizing generative AI. These can be in the form of:
- Build Exercise: You’re given a prod spec and must use AI to build a functioning app.
- Debugging / Comprehension: You’re dropped into a likely unfamiliar IDE. You’re then asked to debug failing test cases and asked to add new features.
AI Reality Check
If this type of interview shows up on your loop, be sure to get specifics from your recruiter on the interview set up and expectations.
Which One Are You Facing?
The biggest mistake candidates make is preparing for Type 1 when the company actually tests for Type 3. When you get that initial call from the recruiter, don’t be afraid to ask: “Can you tell me more about the format of the coding session? Is it algorithmic, or is it more focused on refactoring and code review?”
Need to sharpen your skills? Knowing the format is half the battle; performing in it is the other half. If you’d like to set up a mock interview—whether it’s a traditional coding challenge or a deep-dive code review—let’s chat. We can simulate the exact environment you’re walking into.
I've spent over 25 years navigating the software engineering landscape—scaling teams, fostering well-being, and mentoring the next generation of technical leaders. After 6.5 years at Google, I built a coaching practice to help talented engineers and techies unlock their potential and build the momentum that transforms careers. I'm the coach who meets you at your pivotal moments and helps you make the most of them.