Onsite Interview Guide 2026

The Modern Onsite: Virtual vs In-Person

Most FAANG and top unicorn onsites are virtual in 2026. The structure is the same as in-person: 4-6 rounds with 5-10 minute breaks between. The key logistical difference: your home environment is now part of the interview. A quiet room, stable internet, working audio, and a second monitor for code sharing are not optional. Test all of these the night before.

For the few companies still doing in-person onsites: arrive 15-20 minutes early, bring water, and identify restrooms before your loop starts. These are not minor details. Physical discomfort compounds over 6 hours.

Energy Management Across Rounds

An onsite loop runs 4-6 hours total. Cognitive performance degrades across that window. Candidates consistently underperform in later rounds not because of skill gaps but because of fatigue. Treat the onsite like an athletic event: manage your energy budget, not just your preparation.

Between rounds: do not review notes or try to mentally replay what you said in the previous round. It is wasted energy and increases anxiety. Stand up, move, eat something small, and drink water. Come into each round fresh, not optimizing the previous one.

The night before: do not grind LeetCode. You will not learn a new skill the night before an interview. You will only exhaust yourself. Light review of your behavioral stories (30 minutes maximum) and early sleep is the correct preparation.

What a Bad Round Means

Every experienced candidate has had a bad round in a loop they passed. Debriefs aggregate signal across all rounds. One round where you struggled does not mean you are done. What matters is the overall packet.

The mental mistake to avoid: treating a bad round as if the interview is over. Candidates who mentally give up after a difficult round perform worse in subsequent rounds, making a recoverable situation unrecoverable. Treat each round as independent. Even if you are certain the previous round was weak, perform at full effort in the next one.

Questions to Ask Your Interviewers

Most rounds end with 5 minutes for your questions. Not asking questions, or asking weak questions, is a missed signal. Good questions: "What does a strong first 90 days look like for someone in this role?", "What technical problems does the team face that I'd be working on?", "What's the biggest challenge the team is currently navigating?"

Questions that score poorly: "What's the culture like?" (too generic), "Do you use Agile?" (too trivial), "How much do you travel?" (signals wrong priorities). Ask about the actual work and the team's current challenges. Interviewers remember candidates who asked thoughtful questions.

Energy Management Across a Full Loop

A 5-round onsite is a marathon. Rounds 4 and 5 are where most candidates underperform due to cognitive fatigue, not technical gaps. The mitigation is mechanical: brief walks between rounds, water and a small snack at the 2-hour mark, and a hard mental reset between rounds (you do not carry the previous round's perceived performance forward).

Reports on LeakCode tag "tired in last round" as a consistent failure mode at virtual on-sites specifically. The compressed format (no physical breaks, screen fatigue, no walking between rooms) compounds the cognitive load. Schedule 5-10 minute breaks deliberately even if the interview schedule does not list them; ask the recruiter for these explicitly when the schedule is being set.

What Each Interviewer Is Calibrating

Each round in an onsite is calibrated against a specific rubric dimension. Coding rounds: correctness, code quality, complexity reasoning, communication. System design rounds: requirements clarification, design judgment, operational concerns. Behavioral rounds: scope of impact, leadership, ambiguity tolerance, conflict navigation.

Knowing the rubric lets you allocate effort within each round. The coding round is not graded on "did you solve it"; it's graded on the four dimensions above with the solve binary as one input. Strong candidates explicitly hit each dimension out loud (state complexity, ask clarifying questions, verbalize edge cases). Weak candidates hyperfocus on the binary outcome and miss the other dimensions.

Hiring Committee vs Manager Decision

Companies vary on how the loop feedback gets aggregated into a decision. Google uses a hiring committee (HC) that does not see the candidate; they review written feedback packets and decide. Amazon uses a bar raiser who explicitly weights the loop. Meta and Microsoft are more manager-driven with HM having significant veto power.

For HC-driven processes (Google), your interviewers' written feedback matters as much as the in-room performance. Make their feedback easy to write strongly: name your approach out loud, state complexity, articulate trade-offs explicitly. For HM-driven processes (Microsoft, Meta), the rapport with the hiring manager during their round is disproportionately important. Practical implication: research the company's process and adapt your approach.