What the rounds are like
| Stage | What people usually describe |
|---|---|
| Online assessment | 90 minutes, one problem in four levels, like an in-memory store or a file system |
| Phone screen | Two problems in 45 minutes |
| Onsite coding | One or two rounds, usually two problems each |
| AI-enabled coding | 60 minutes in a small existing codebase with an AI assistant. Newer loops swap it in for one coding round |
Candidates say you get about 35 to 40 usable minutes for both problems. Practice explaining the approach and complexity quickly, then coding without stopping much. Watch for small twists on the LeetCode version.
What people have been asked
These come up in candidates' own write-ups of their interviews. We wrote the example prompts ourselves, so the wording you get will be different.
Raise a number to a power
Implement exponentiation yourself, including negative exponents.
Multiplying n times is too slow, so interviewers expect you to halve the exponent each step. Watch edge cases like 0 to a negative power.
How it might be asked
Implement pow(x, n) without using a built-in power function. n can be negative or zero.
Key concepts to know
Try this Interview
Fix unbalanced parentheses
Remove the fewest parentheses to make a string valid.
Track which brackets are unmatched as you go, then remove exactly those. A common follow-up: do it without the stack, in two passes.
How it might be asked
Given a string with letters and parentheses, remove the fewest parentheses so the result is valid, and return any valid result.
Key concepts to know
Try this Interview
Vertical order of a tree
Return a binary tree's values column by column, top to bottom.
BFS keeps top-to-bottom order for free, and a hash map groups the columns. Follow-ups: skip sorting the columns by tracking the min and max, and why BFS over DFS.
How it might be asked
Given a binary tree, return its values in vertical order: column by column from left to right, and top to bottom within each column.
Key concepts to know
Try this Interview
Make the largest island
Flip one water cell to land and return the biggest island you can make.
Checking every flip from scratch times out. Work out each island's size once and reuse it.
How it might be asked
You're given a grid of 0s and 1s. You can change at most one 0 to a 1. Return the size of the largest island you can end up with.
Key concepts to know
Try this Interview
How to approach a problem
- Say it back and ask questions. How big is the input? What are the edge cases? What should you return?
- Start with brute force. Say it out loud with its complexity so you've got something to beat.
- Figure out what it really is. Is this secretly a graph? Would sorting help? Do the same subproblems keep coming up?
- Write it and test it. Walk through an example, then try the weird cases.
Key concepts to know
- Trees and recursionSolve tree problems by deciding what each node returns.
- Graph traversal and shortest pathsBFS, DFS and Dijkstra, and when to use each.
- Hash maps and sliding windowsSwap nested loops for a single pass.
- Binary searchCut the search in half, even when you're searching for an answer.
- Heaps and priority queuesAlways have the smallest or largest item ready.
- Dynamic programmingSolve each subproblem once and reuse the answer.
- Stacks and monotonic stacksMatch brackets, parse things, and find the next bigger value in one pass.
- Union-findKeep track of connected groups as links keep arriving.
Sources
We pulled this together from candidates' own write-ups of their interviews. They don't always agree, and things change over time.
- LeetCode Discuss: Meta E5 offer, Menlo Park (2024)
- LeetCode Discuss: Meta E4 onsite passed (Feb 2025)
- LeetCode Discuss: Meta E5 onsite, Bangalore (2025)
- LeetCode Discuss: Meta product E4, news feed design (2025)
- LeetCode Discuss: Meta full loop with AI-enabled round (Oct 2025)
- LeetCode Discuss: Meta SDE2, AI-enabled round and product architecture (Nov 2025)
- LeetCode Discuss: Meta E4 US offer, DoorDash design (Dec 2025)
- LeetCode Discuss: Meta E4 Bangalore, AI coding and product architecture (2026)
Practice Loop is an independent practice tool and isn't affiliated with or endorsed by Meta.