Google SWE Interview Playbook

Google’s interview is coding-heavy and committee-decided. You will spend most of the onsite on whiteboard-style coding problems, and you will never meet the people who actually decide whether you get an offer. Both of those facts change how you should prep.

The loop

Standard L4/L5 SWE onsite is five rounds:

1. Coding ×3 — 45 minutes each. LeetCode medium is the floor; hard problems are common at L5+. You are expected to state complexity before and after your solution, handle edge cases without being prompted, and discuss tradeoffs between approaches.
2. System design — only for L5+. Senior signal is the entire point of this round. Ambiguous prompt, you drive.
3. Googleyness & leadership — behavioral, scoped to collaboration, ambiguity, and feedback. One round.
4. General cognitive ability / domain — varies by team. Sometimes a second design round, sometimes a deeper domain dive.

After your loop, interviewers write long packets. The hiring committee reads packets plus resume, calibrates against the level bar, and votes. You do not meet them. Offers are made after HC approval plus team matching.

The coding bar

Google’s coding bar is specific: clean solution, correct complexity, tested. The difference between L3 and L4 on the same problem is usually not the solution — it is whether you stated the complexity up front, validated your approach with an example before coding, and handled the obvious edge cases without being asked. The difference between L4 and L5 is whether, when stuck, you narrated your debugging and converged on a fix — or froze.

Two anti-patterns that fail this bar reliably:

1. Silent coding. You write for ten minutes without narrating. Interviewer cannot score your reasoning, so they score “no signal.” No signal is a no hire.
2. Skipping complexity. You finish the code and do not mention time/space. The interviewer has to ask. That knocks your coding score down a half-level by itself.

The prep correction: in every mock, narrate the approach, state the complexity before writing, and give three edge cases without being prompted. That’s the muscle the Google bar is checking.

Leveling signals

Google’s levels are well-defined, and the interviewer is explicitly asked to score you against a level bar. A rough shorthand:

• L3 — correctness, complexity rigor. You solve the problem, you say why your solution is O(n log n), you test it.
• L4 — tradeoffs. Multiple valid approaches, you pick one and justify it against the alternatives. You catch your own bugs.
• L5 — scope and ambiguity. Design round is graded on how you clarify requirements, handle unknowns, pick what to build first, and defend those choices. Senior signal = you drive the conversation, not the interviewer.
• L6+ — organizational scope, cross-team technical leadership, mentorship stories. Coding rounds shrink, leadership rounds expand.

If you cannot clearly articulate a project you led that fits the level you are targeting, you are going to underperform the level bar regardless of coding ability.

The hiring committee calibration

Most candidates fail not in a single round but in signal consistency. HC wants to see the same level across your packet — strong coding, strong design, strong behavioral. One weak round that you “made up for” with a great one often scores below a candidate who was uniformly solid. The bar is not “average above the line.” It is “nothing meaningfully below the line.”

Pair this with a mock interview graded against Google’s coding-and-complexity rubric, and cross-reference Amazon’s LP-driven loop and Meta’s product-sense bar to see why the same solution earns different scores.