grad school advice

Thoughts drawn from my experience doing a CS PhD at a top-4 school around 2010-2016. They may be somewhat applicable to PhDs in other areas (and even to junior-level positions in industry) but you'll have to judge that yourself.

• Only go to grad school if you can't not go to grad school.
  • These days there are great opportunities to do CS research in industry (residency programs and engineering jobs in the research organizations at Google, Meta, and most of the other big tech companies), where you'll get paid a ton, be exposed to real-world problems, develop collaborative skills, likely have a healthier culture, and still get to learn from very smart people and often publish research papers. I suggest trying this route first, (especially if you don't know exactly what you'd want to do in a PhD) and only moving on to a PhD once you're sure that nothing else will satisfy you.
• A PhD is an apprenticeship. You're not choosing a school, you're choosing an advisor. To first order, your school choice is important only insofar as it constrains your choice of advisor (though second-order considerations like location may still be quite important to you).
  • Do you personally vibe with them? That's hugely important.
  • Ask their current students for frank advice. Do they recommend working with this person? What are their pluses and minuses?
  • Working with a good advisor at a non-prestigious school is infinitely better for your happiness and career prospects than a bad (or bad fit) advisor at a prestigious school.
  • You'll get a lot more mentorship from being the first or second student of a bright young professor than from being the Nth student of a famous senior professor. It can be worth having a senior person as a secondary advisor (see below), but if you have to choose, prefer the young advisor who'll be more invested in you.
  • As with all relationships, you'll need to be comfortable being vulnerable with your advisor: when you have concerns or when things aren't going well, you need to let them know and give them the chance to help. During your initial conversations, consider admitting some insecurity and gauge how they respond. If they're not supportive (or if you don't feel comfortable doing this because you don't expect them to be supportive), it's better to realize this sooner rather than later.
• Don't work on research you don't love. The benefit of grad school is that it frees you to do passionate, purposeful, self-motivated work. Your passion may ebb and flow; there will be a slogging phase to every project, and not all projects will succeed, but there needs to be something about your work that excites you, such that you're genuinely curious about the questions you're exploring and that you'll be proud if and when your work does manage to answer them. Evaluate your situation every 6-12 months; if you can't locate that kernel of passion, then you're exposing yourself to the downsides of academia without most of the upsides. At that point you don't have much to lose; you should be willing to take great risk (switching projects / advisors) to try to rectify the situation, and don't be afraid to leave if it doesn't work out.
• Set achievable goals: it's much, much better to do ten smallish projects (ideally building on each other so that they add up to something bigger) than to attempt a big all-or-nothing project. Each project you finish adds to your CV, builds your confidence, introduces you to new people and ideas, and gives you experience with the end-to-end lifecycle of a research project. I got bogged down working on a single project for my whole PhD, and as a result, I got relatively little experience with the final stages of the research process (writing up the work and presenting it) and the initial stages (scoping and formulating a new project), both of which are incredibly important to a successful research career.
• Your first several projects may not meet your standards; that's fine and normal. The solution is to always produce: you'll get better as you go.
• If you ultimately want an academic or otherwise research-focused job, you must prioritize finding and join your research community. This is probably the aspect of growing as a researcher that I screwed up the most.
• Many of the most successful students I knew had multiple advisors. This requires (and also encourages) a pretty high level of self-motivation and confidence in your research identity, since you have to reconcile potentially conflicting advice and open up your work to criticism from multiple perspectives. But such criticism can be incredibly valuable; it tends to ensure that you work on things that are seen as broadly meaningful rather than one person's pet project. And getting exposure to multiple role models allows you to learn from the best of each.
• Research internships are a great way to get temporary exposure to a different advisor and a new set of problems, and also to industry salaries. Don't do irrelevant internships just for the money, but if you have the opportunity to work with a researcher in your area whose work you respect, you should generally jump at it.