Do PhD admissions (CS) in the US work on two levels (general decisions and PI decisions)?

Question

Going through the computer science PhD applications process for US schools, I've heard two conflicting types of advice/anecdotes:

1. While they have a certain broad area in mind before starting their PhD studies, PhD students gravitate towards faculty members working on specific problems that interest them during their beginning years.

2. During the admissions process, potential PhD students are interviewed by the professors they want to work with. In this case, their admission seems to depend on whether that specific professor is recruiting PhD students for that year or not.

This gave rise to a few questions:

Are there two implicitly different tracks to being admitted as a PhD student? This can also be thought of as the difference between applying straight from undergrad (will take courses during PhD) and with an MS degree (can skip some courses).

In other words, are students with different levels of "research-readiness" evaluated differently?

Answer 1

In most fields in the US including CS (especially theoretical CS), you are admitted to a "program", essentially to a department. The decision about an advisor comes later. If you enter with a BA/BS then it can be much later (years). The first task is to take whatever courses are necessary to get you through prelim/qualifying examinations, which are all on advanced topics. Even some MS entries may need some courses and need to pass quals. About that time (bit earlier or later) you make arrangements with an advisor for your dissertation work.

In some scientific fields the PI decision may need to come earlier, but, perhaps, only in general - with a group or lab, rather than an individual.

But, yes, they are usually separate decisions. In some places in Europe it is quite different and "admission" to a program is the same thing as being "hired" by a PI. It isn't like that in US. A few situations are more like the European model, with a grant funded PI who is able to "admit" doctoral students into their lab or working group. But this is rather rare in US and most funding and admissions is at the department level. See the answer of Bryan Krause for more on the alternative model. Applied fields might be more like this, including Applied CS, since a PI might have funding to work on a specific problem.

Until you pass quals in US, all agreements with PIs/advisors are tentative and conditional.

---

Funding: On the standard US model, most students are funded as TAs or (less frequently) RAs. This includes forgiveness of tuition fees as well as a modest stipend sufficient to live on. Getting such a position is, again, usually up to the department, not the individual professor. Large departments that teach undergraduate courses (math, CS, ...) need a lot of TAs.

A few students are funded by employers, but this seems to be rare currently. It was more prevalent in some places in the past. And, a few international students are also funded by home governments.

Answer 2

To add to @Buffy's answer, a lot depends on where funding comes in your research area.

If a department has funding for students, through assigning teaching positions or training grants that fund students directly, students will be admitted per program. It is still up to professors whether they want to take on particular students as their mentors, which may depend on all sorts of things - how many students they have already, whether they think the student is a good fit, etc.

If individual professors have funding for students, then this can present an additional hurdle or an additional path to admission. Professors may be limited by the funds they have to pay graduate students, so a student can't work with a professor that has no available funds even if the professor wants them (unless the department can pick up the slack). Additionally, there may be students that a professor is willing to fund but who miss the cut for department-level funding. These students may be admitted directly into that professor's lab (typically they still need approval from the department, but it's common for the "number of students we would love to have here" exceeds the "number of students we can afford to have here"; it's students that fall in the first but not second category that this applies to). In my own area of study (neuroscience) this path seems especially common for international students, because some of the other funding sources for training students are limited to domestic applicants.

Programs you apply to should discuss these possibilities.