The grad-school interview checks exactly three things

Three checks: you did it, you mean it, your English survives a seminar

Every question maps to one of three checks. One: is the research on this resume really yours? That is why they dig into your projects. Two: do you know why this specific program, beyond its ranking? Three: can your English carry an hour of seminar debate with classmates?

You can prepare for all three. You cannot fake any of them.

The invitation already tells you which interview this is

Two very different rounds hide behind the same congratulations line. If the mail came from one professor, from their own address, with a link to their own calendar and a thirty- to sixty-minute slot, that is a one-on-one. PhD admissions and lab-based master programs mostly look like this. If it came from the department or the admissions office, with slots they have already blocked out for you to choose from, and the words faculty panel or admissions committee somewhere in the body, or three names listed at once, that is the committee round. Fifteen to twenty-five minutes, start to finish.

For the one-on-one, start from the professor’s situation. They are choosing the person who will sit in their lab for the next two to five years, quite possibly paid out of their own grant. They are not scoring you. They are estimating how much of their week you will cost them. So the questions go deep, and they will cut in halfway through an answer to ask why you did not use the other method. That is not hostility. That is them simulating a lab meeting with you in it.

When you get cut off, do not rewind and replay the script. Answer the question they just asked, stop, and let them decide whether to go back to the original thread. The usual way to lose a one-on-one is to be interrupted mid-recital and then spend ten seconds hunting for where you were.

• If the mail does not say, ask: how many faculty, how long, is a presentation expected, camera on or off. Asking costs you nothing. Finding out on camera that they wanted ten minutes of slides costs you the interview.
• If a professor attached their lab page or one or two papers, that is not politeness. That is the assigned reading. They expect you to have read it, and they will ask what you made of it.
• If they hand you on to a PhD student in the lab for a second conversation, that is not a demotion. That student is the one who would share a bench with you, and in many labs their opinion carries real weight.

The committee round: five people, twenty minutes, nobody keeping the thread

The physical facts of a committee interview: four or five faculty, twenty minutes, a dozen candidates to get through that day, and most of the people in the room are not in your subfield. The deep discussion you imagined does not happen. What happens is one or two questions each, then the next person, with nobody in the room responsible for joining your answers into a picture.

So the strategy inverts. A one-on-one rewards drilling down. A committee round is about getting a room of strangers to remember three sentences: what you have done, what you want to do, why this department. Whatever they ask, the answer has to end up back at one of those three.

Pitch the technical detail at the least specialized person in the room. That does not mean flattening it into something false. It means changing the layer: say what problem you were solving and why it was hard before you name the method. If you finish and nobody picks it up, they usually did not follow you, and they will not ask a second time for your sake.

• Look at whoever asked, sweep the others once mid-answer, come back. Talking to one face for twenty minutes tells the other four this has nothing to do with them.
• One person will say nothing all round. That is not boredom. They may be chairing and watching the clock, or they may be your future advisor, watching. Do not write them off because they gave you nothing back.
• Asked something you genuinely cannot answer, say you have not worked on that, then bridge to the nearest thing you have. Bluffing is worse: at least one person in that room was scheduled precisely because that question is theirs.

“Tell me about your research” is a question about decisions

This one is close to guaranteed, and most people answer it as a chronology: first we did this, then we did that. The professor is listening for decisions. Why that problem, why that method, and what you changed when it stopped working.

• The two-minute outline: the question (30s) → why it was hard (20s) → your approach and your own part (40s) → the result (20s) → limitations and next step (10s).
• Do not skip the limitations. An applicant who can say where their own work is weak sounds like a colleague. One who only lists strengths sounds like a salesperson.
• You must be able to defend every number on the resume (sample size, accuracy, ranking). A number you cannot defend is far worse than no number.

“What do you want to work on” has to land at the level of a question

Both of the obvious answers fail. Too broad: “I want to work on AI for healthcare.” Thousands of applicants wrote that sentence this year, and it leaves the professor with no idea what you would actually do on a Tuesday. Too fixed: “I will compare these two methods on that dataset.” It sounds like you start Monday and do not intend to be supervised.

The shape that passes is a narrow question with the method still open. Some effect does not hold under some condition, and you suspect one or two ways in, depending on whether the data can be got. Narrow enough to sound like a real question, then one sentence saying you are still working out how to attack it.

Still working it out is not weakness. Nearly everyone changes their plan in the first year, and the professor knows that better than you do. What they are checking is whether you can find a problem yourself and judge whether it is worth doing. People who already have the answer do not usually need the degree.

• Prepare two directions, not one. When they say the lab has moved off that line, you need a second sentence, not three seconds of silence.
• You must be able to say how your question attaches to something the department already has: a lab, an instrument, a long-running dataset, a method one of them built. A topic that attaches to nothing is not their topic, however well it is written.
• If your statement of purpose already names a plan, the spoken version has to match it. Changing your mind is fine as long as you say why. Otherwise the committee starts wondering who wrote the statement.

The passing bar for “why this program”: two names

“Great ranking and resources” scores zero, because it fits any school. The bar: name two professors in the department, say what they are working on lately, and connect it to what you want to do. One afternoon reading their recent paper abstracts covers it.

The bonus move links your project to one of theirs: “I worked on X, and Prof. Y is extending X toward Z. That is the reason I applied.” One sentence covers checks one and two at once.

They will ask about your weak spot, so draft the answer now

• A GPA dip: give one honest cause plus later proof, as in “that year was low, and the four semesters after were 3.9 with all A’s in the field.” Saying nothing reads worse than the dip.
• Changing fields: do not apologize. Say what the old field gives you (methods, a point of view). Then say what you did to earn the new one (courses, projects, proof of self-study).
• A gap, or returning after years of work: committees don’t mind your age. They test whether your reason can survive two years of hard, slow work.
• In the editor, switch the industry style to Grad School after generating, and run the mock interview. The questions come from the admissions side, including the one aimed at your weak spot.

If your English is shaky, spend the effort on listening

Interviews rarely collapse over grammar. They collapse because the candidate did not catch the question and answered a different one. Faculty tolerance for an accent, a wrong tense, a filler word is very high. Tolerance for an answer that misses the question is very low, because that looks exactly like someone who will not follow a seminar. So prepare in that order: listening first, speaking second.

The method is dull and it works. Find recorded talks by the professors you are applying to, departmental seminars, podcasts in your subfield, and listen until their speed and their habitual phrases stop surprising you. Academic English has a narrow surface. The words that actually recur in your field number in the dozens, which makes it far easier to prepare than everyday English.

For speaking, complete sentences and a pace you can control are enough. Slow beats wrong. Short beats long. Memorize two lines for the moments you get stuck: one that asks them to confirm you understood the question, one that asks for a few seconds to think. Using either costs you nothing. Talking over a question you did not hear costs you the round.

• Writing ahead of time to say you are not a native speaker and would appreciate a slower pace is a normal request in most departments, and people generally do slow down.
• Drill the two-minute version of your research until it needs no thought. It is the one passage worth memorizing, because it is certain to come up and because a clean opening steadies everything after it.
• General English interview practice is its own subject and its own article. One thing to add here: the academic vocabulary barely overlaps with the business one, so do not practise on the wrong material.

Money, and “where else did you apply”

“Where else are you applying” comes up almost every time, and the person asking genuinely wants the answer. They are estimating their own yield: if we make an offer, will this person come. The worst answer is “only here”, because nobody applies to one program, and that sentence files you under dishonest.

Name two or three comparable places, then say what they have in common. If you applied to the same kind of lab everywhere, the list becomes a witness for you: it shows you know what you want. You do not owe them a complete list. It only has to look like it came from a logic rather than off a ranking table.

Raise funding yourself instead of waiting for them to mention it. For a PhD: how many years are guaranteed, RA or TA, are summers included, what happens if the advisor’s grant is not renewed. For a master: are there TA slots, does the department have merit scholarships, when do decisions go out. These are the terms of a contract you sign for two to five years. The applicant too polite to ask is the one who finds out at registration that the tuition is entirely theirs.

• Asked “if we admit you, will you come”, do not overcommit to please them. Honest without closing the door: this is my first choice today, and I am waiting on one thing. Promise and then renege, and you will find the field is much smaller than you thought.
• Do not push one offer against another to move the number. This is not a salary negotiation, that posture reads badly in an academic room, and the person across from you usually has no authority to change the figure anyway.
• If they volunteer that funding is available, pin down the years and the conditions in the moment, and ask them to put it in writing afterwards. A verbal promise does not survive a change of staff.

The last five minutes: your turn

• Ask how the program actually runs: first-year milestones, how rotations work, when students usually settle on a thesis topic.
• For a master’s: ask plainly where graduates end up. For a PhD: ask how many years of funding are guaranteed. These are contract terms for years of your life, and asking about them is the professional move.
• Never ask what the program homepage answers. That question only confesses you did not read it.

After the interview: which “keep in touch” is real

What they say at the end comes in two kinds. The first is “write to me if you have questions”, or “good luck with your applications”. That is courtesy. It carries no information, and it is not a hint about the outcome. The second kind has a concrete next step inside it: I will tell you after the committee meets next week, send me that code, I will introduce you to a student in the lab. Take the second kind literally, and do it inside the window they implied.

Within a day, send a short note to every person you spoke to. The academic version of the follow-up note is not the job-hunting version: no declarations of enthusiasm, no restating how badly you want to come. One paragraph. Thank them for the time, name one specific thing you discussed, add the point you did not finish making or a link. They may not reply. The note sits in the inbox anyway, and on the day of the meeting they see your name.

Then go quiet. Departments usually give a timeline. A short note once that timeline has passed is normal. Chasing weekly inside it is not persistence, it is a nuisance, and it is remembered.

• A waitlist is not a consolation prize. It means you are in as soon as someone ahead of you declines. This is one of the few moments where a short note saying the department is still your first choice actually does something, because when they go back to the list they care most about who will definitely come.
• After a rejection you can write to thank them for their time and ask what to strengthen before applying again. Some professors answer in detail, and those few lines are worth more than any paid advice.
• After an admission, do not send only thanks. Ask what to read, what to take, which tool to install and practise on in the three months before you arrive. People who start early have a very different first year.