Virtual interviews vs in person: what actually changes?

You have probably been told a virtual interview is the same conversation, just on a screen. That is half true, and the half that is wrong is the half that costs offers. The differences come down to three things: how much of you the interviewer can actually perceive, who controls the environment, and how turn-taking works.

• Narrower signal. A camera captures your face and shoulders. Posture, hand gestures, and the energy you bring into a room are mostly gone. What is left carries more weight, so vocal energy and facial expression have to do more work than they would across a table.
• Turn-taking is harder. Network delay means you and the interviewer will talk over each other. Waiting a beat after they finish is not awkward on video; it is the correct behavior, and it prevents the collision that makes both people apologize.
• Silence stretches. A three second thinking pause across a table reads as consideration. The same pause on video, with no shared room to absorb it, can read as being stuck. Saying what you are doing ("let me think through the data model first") converts dead air into visible reasoning.
• Your environment is on the record. Lighting, background, and mic quality become part of the impression whether you intend them to or not. None of this makes you a better engineer, and all of it is judged.
• Notes are possible. Off camera, you can keep a page of prepared material in view. This is the one structural advantage the virtual format hands you.

Platform specifics differ more than people expect. Round-ups of Zoom, Google Meet, and Microsoft Teams behavior live in the Interview Types topic.

The onsite round is not just the virtual round with a commute attached. It measures things the video call structurally cannot reach.

• Informal assessment. The walk to the room, the lunch, the engineer who chats with you while the next interviewer sets up. These are unscheduled and frequently reported back. You are being evaluated in the gaps, not only in the slots.
• Whiteboard thinking. Standing at a board changes how you explain a system. You point, you erase, you redraw. Interviewers read your process from the physical trace of it in a way a shared screen flattens.
• Room reading. In a panel, you can see which member disengaged and adjust in real time. On a grid of tiles, half of that is invisible.
• Fatigue and stamina. A five hour onsite tests whether your fourth answer is as sharp as your first. Virtual loops are usually shorter and more fragmented.

There is also a plain scheduling asymmetry. Companies spend money to fly candidates in, so an onsite invitation generally means you have already cleared a bar. Virtual rounds screen; onsite rounds usually decide.

The content preparation is shared. Your stories, your project depth, and your fundamentals do not change with the medium. The delivery preparation splits.

For virtual, do a real technical dry run: the actual laptop, the actual headset, the actual room, at the actual time of day so the light matches. Check that your mic does not clip and that your camera is at eye level rather than looking up at you. Practice answering while looking at the lens instead of at the interviewer's face on screen, which feels wrong and looks right.

For in-person, rehearse standing up and talking through a design out loud, and plan the logistics so travel is not eating the attention you need for the interview.

Consider a backend engineer interviewing for an L5 role at a public cloud vendor. Her first two rounds are on video, so she pins her three strongest project narratives beside the camera and practices the four second pause that keeps her from stepping on the interviewer. The onsite is a whiteboard system design, so she practices that one standing at a wall with a marker, not sitting at a laptop. Same material, two different rehearsals.

Either way, the cheapest rehearsal is a full spoken run rather than reading notes silently. Timed practice with an AI interviewer is on the mock interview page.

The format decides whether real-time assistance is even a possibility. In a room with a table and a whiteboard, there is no screen to put anything on, so live assistance is out and preparation is everything.

On a virtual call, a desktop assistant like SubcueAI runs as a native app on macOS or Windows, captures both sides of the conversation, and shows suggestions in a local floating overlay. It does not join the meeting as a participant and does not inject anything into the browser during a live interview. A separate Chrome extension exists for mock-interview practice only.

The honest limits matter more than the capability. If you share your screen, the overlay is on the screen you shared. If the session is recorded on your side, it is in the recording. Proctored assessments and company-managed devices are outside what any of this is designed for, and treating them otherwise is how people get into trouble. Aaron Cao, founder of SubcueAI, built the overlay to sit outside the meeting client specifically because a tool that pretends those limits do not exist is a tool that fails a candidate at the worst possible moment.

The full picture of what is and is not visible is in the Detectability & Privacy topic, and setup steps are on the tutorial page.