Google Coding Interview: What Gets Asked

Google's coding interview is built from a small set of data-structures-and-algorithms patterns: graphs and grids lead, followed by arrays and hashing, two pointers and sliding windows, trees, dynamic programming, and heaps. Questions run at LeetCode medium-to-hard, take place in a plain Google Doc with no IDE, and are scored as much on how you reason as on whether you reach the answer. The most reliable prep is to drill the recurring patterns, not to grind a random problem list.

Three of your four onsite rounds at L3 and L4 are pure coding. At L5 it's three of five. So more than half of what stands between you and an offer is one thing: a blank editor and a problem you've never seen. Here's the part the reputation gets wrong. Google is predictable. The same handful of patterns come around again and again, and you can drill every one of them before you walk in.

How a coding round runs

Each round is 45 minutes in a shared Google Doc. No IDE, no autocomplete, and expect no way to run your code, though some candidates now get a basic editor with syntax highlighting. You check correctness in your head and on paper as you go. A rough split holds up: five to ten minutes understanding the problem and pinning down examples, then twenty-five to thirty-five writing and talking through the code. Leave the last few minutes to trace your solution on a concrete input. That trace is where strong candidates lose points, by skipping it.

One recent change matters. In May 2026 Google started piloting an AI-assisted format for some junior and mid-level roles. Instead of writing an algorithm cold, you read and debug an existing codebase with Gemini on hand. The pilot is narrow. Most candidates still get the classic format, but ask your recruiter which version your loop uses.

What the bar really is

The patterns that matter

Google won't hand you a question list. But the mix isn't a mystery to anyone who has sat the loop, and it's lopsided. A small set of patterns covers most of what gets asked. That settles the order you study them in.

Graphs and grids come out on top, with the array-and-hashing fundamentals and the two-pointer family close behind, and the rest spread out behind them. These are the pattern families we teach, and this section walks them in roughly that order. Each lesson opens with a short read on how Google frames the pattern and the variants they favor. Then it hands you one to four of the most-asked problems as runnable exercises with worked solutions. The full algorithm walkthrough for each lives in the Foundations course, linked from every lesson, so these pages can stay on what's specific to Google.

Common questions