How (I Think) I Got into Harvard & Columbia

This is an FAQ based on my graduate admissions processes and experiences as a student across multiple Ivy League schools.

I was a student at Harvard (master’s) from 2021-2023 and now at Columbia (PhD), so I’ve gotten a decent taste for both types of programs (in engineering and statistics, respectively).

Is everything in this FAQ anecdotal and therefore potentially inapplicable/horrible advice for your own best path in graduate admissions?

Yes, all advice is. Please don’t use this as a handbook.

Are my grades high enough for an elite school?

Some people will say grades don’t matter, and if you have amazing rec letters and publications in top venues (for me, that’s NeurIPS, ICML, ICLR, etc.) with bad grades, then they’re probably right.

Odds are, however, you don’t. This sort of student is quite rare. PhD students often purposefully get lower grades to work on their research, but undergrads usually don’t. If they get bad grades, it’s either because they had a lot going on in their personal lives or they straight up didn’t understand the material. And that can be fine for one-offs! It happened to me. Just don’t make it a habit.

If you were going to procrastinate on classes with other tasks, it’s probably better to do that with interesting side projects with programming components where you have a chance to make something cool/useful, because an undergrad is unlikely to have the toolkit to crack really niche or cutting edge research problems in stats or ML unless they’ve been doing hardcore research for 2-3 years and get lucky with stellar mentorship.

My personal advice here is to take as many hard classes as you possibly can in undergrad. You’ll get better at thinking even if the ramp up is difficult. Lower grades with harder courses are more meaningful (and better for you) than loading up on easy A’s. Eat your vegetables. If you do that, good grades will become a secondary effect of your habits.

But short answer? You’re probably overthinking it. A mixture of A’s and B’s aren’t gonna kill you in the app process. I had two B-‘s in physics and survived.

C’s and D’s in core classes will be a major red flag, however.

Most important app components for a master’s vs. PhD?

Master’s? Have a lot of money! No, but seriously? Grades and an interesting story. Master’s programs are mainly there to fund other programs and generate good word of mouth for the university by getting their graduates into big tech firms or hot startups. It’s a brand-for-brand exchange. Some people are pretty cynical about this, but at an elite university like Harvard I actually thought it was mostly a fair deal.

Master’s admissions folks want to make sure you’ll finish the program and get a fancy gig so they can brag about you in their post-grad stats. But the people most likely to do that are the ones who are already good at gaming systems by getting stellar grades, generating engagement or online self-promotion, and networking. Research-oriented master’s students like myself were a minority of Harvard SEAS, so your master’s essay probably shouldn’t even mention professor names (except maybe one brief sentence).

PhD? I’d say rec letters first, publications second. Those two things are 90% of a PhD application. If your grades are even average but you can kill research, you’re going to have your pick of schools. You can (should) add non-archival projects to your resume, but know that an actual publication in a recognizable journal is infinitely more impressive.

Bottom line, though, is that academia is still a who-knows-who game. Would you trust a competent friend/colleague to recommend you a product you’re gonna use every day for 5 years more than a complete stranger? I would.

Where to apply?

One underrated aspect of grad school is being able to, you know, EXIST in that location! Looking back, there were some schools I got into in certain cities that I would never move to in retrospect. These applications were a waste of time.

Don’t just focus on the school. By now, having presumably lived on your own for four years (or more) you’ve developed some personal preferences. Don’t dismiss these preferences! Grad school is miserable enough without hating where you live, too. You may grow to hate it, but at least don’t start off like that.

After that, the rest is a voodoo mixture of “where I think I can get in” with “where would afford me the most opportunity.” Nobody knows this exact formula, and you’ll probably mess it up. It’s just how it is. That’s why I applied to 10 schools, to average out the noise.

If your rec writers have colleagues at other schools, you may want to base a lot of your search off of that.

What is NOT on the application that I should be doing?

Reach out to professors via email to ask if they’re considering new students in the upcoming cycle UNLESS they say not to on their website (doing otherwise may piss them off).

Do this in August and September before the crowds start blowing up their inboxes.

Make these emails short and to the point so you get a response. For the love of God, don’t waste a sentence on “My name is.”

Be personable.

Use whitespace generuously.

Make each sentence its own line/paragraph and use no more than 3-4 clean sentences. Choose these sentences carefully and maximize each for impact!

With this strategy I was able to get a response with over 75% of my emails. Though the answers varied in utility and effort, a bad response is better than no response.

Anything you regret from your application cycles?

I regret optimizing for prestige. Research is hard, frustrating, and always slower than you want. Even if prestige might afford you more immediate opportunity, prestige means much less after you get in (i.e. the rest of your life) and much less than cold, hard skills. Doing good work is a much more satisfying reward, and as you transition out of undergrad, you’ll realize just how many animals there are in “less-prestigious” programs that are more skilled than you in other important ways.

When I first got to Cambridge, I was amazed at how many smart people I was suddenly surrounded by. But I came to know many students at both Harvard and MIT who were working under people who didn’t care if they were hit by a bus, and besides being completely miserable and constantly stressed, a lot of them didn’t even finish their programs!

So don’t optimize for prestige only. Factor it in, but don’t be blinded by it. Look up Youtube videos of your potential PhD advisor. Do they seem nice? Talk to their students. They probably won’t spill ALL the beans, but you might get a certain vibe. Check their websites and cross-check LinkedIn! Do a lot of their students drop out of their lab?

Another way to get around this is by looking at programs where you don’t immediately have to choose an advisor in your first year. Harvard is NOT one of those programs (in 2024, at least). Columbia is! And it didn’t matter anyway, because the stats department is wonderful (shoutout JPC!).