How to Find Professors in Computer Science Accepting PhD Students (2026)

Computer science runs on a different clock than most PhD fields. Funding lines rise and fall with grant cycles, admission usually means joining a specific advisor's group on day one rather than rotating through a few labs first, and the field's own publication culture, arXiv preprints and conference proceedings over journals, changes how you should actually go about confirming a professor is active right now. Generic supervisor-search advice built for biology or the humanities misses all of this.

You're usually applying to a person, not a rotation

In most US computer science PhD programs, you're admitted directly into a specific advisor's research group based on the fit you describe in your application, not placed into a department-wide pool to rotate through multiple labs before choosing one, the way biology and chemistry programs commonly work. That's exactly why naming a specific professor and referencing their actual recent work matters more in a CS application than in fields where the rotation year absorbs some of that risk. A minority of CS departments, often newer interdisciplinary programs, run something closer to a rotation model, so it's worth checking a given program's own admissions page rather than assuming the direct-admit model applies everywhere. In the UK, the pattern is even more pronounced: supervisor-led, project-driven PhDs are the default across fields, and applicants are often expected to identify and make contact with a prospective supervisor before applying at all, sometimes shaping the research proposal together.

Why a professor's funding line can disappear between cycles

A CS professor's capacity to fund a new PhD student through a research assistantship is tied directly to whether they currently hold an active grant, NSF, NSERC's Discovery program, EPSRC, an ERC Starting or Consolidator Grant, DARPA funding, or an industry-sponsored research agreement with a company. Grants typically run three to five years, which means a lab that was fully staffed and funding new students two years ago may have nothing open today, and a lab that looked quiet last cycle may have just landed new funding. This is the single biggest reason a cold email to a CS professor should ask directly and specifically about current funding availability rather than assume it from an old lab page or a department's general reputation for being well-funded.

Grant programs worth knowing by name

In the US, the NSF Graduate Research Fellowship Program funds three years of support, usable within a five-year window, as a portable award the student controls rather than one tied to a specific advisor's grant, which gives a GRFP winner real leverage in choosing an advisor since they arrive already funded. It's restricted to US citizens and permanent residents. In the UK, UKRI sets a centrally mandated minimum PhD stipend applied across all its research councils including EPSRC, which funds most UK computer science PhDs, revised annually, so check the current cycle's published minimum rather than an older figure. In Canada, the previously separate CGS-D and PGS-D doctoral scholarships were merged into a single harmonized Canada Graduate Research Scholarship-Doctoral starting with the 2025-26 competition, a recent enough change that a lot of existing funding-list content hasn't caught up to it yet.

Industry fellowships sit in a different category entirely: the Google PhD Fellowship and the Microsoft Research PhD Fellowship are both nomination-based, your university nominates you, generally in a later year of your PhD, rather than something you apply for directly, and Microsoft's specifically targets third- and fourth-year students in computer science, electrical engineering, and mathematics from a defined list of eligible countries that shifts by cycle. Treat these as a possible honor partway through a funded PhD, not as something to plan your initial supervisor search around. NVIDIA runs a separate Graduate Fellowship worth knowing about too, worth up to $60,000, but it comes with conditions that change how you'd actually use it: it requires an existing faculty advisor to sponsor the application and a mandatory in-person summer internship at NVIDIA, so it's a fellowship to raise with a professor once you're already working with them, not something to apply for independently before you have an advisor.

arXiv and conference proceedings, not just Google Scholar citations

Computer science, and machine learning and theory in particular, is unusual in how much of its research appears on arXiv as a preprint before or alongside formal publication. That makes arXiv a genuinely field-specific tool for confirming a professor is actively working right now, in a way that's far less useful in, say, public health or economics. It's also worth remembering that CS treats top peer-reviewed conferences, NeurIPS, ICML, CVPR, ACL, SIGCOMM, STOC and FOCS among them, as the primary venue of record, roughly the equivalent of a top journal in most other sciences. That means judging whether a professor is currently active is better done by checking their recent conference acceptances and arXiv output together than by relying on total citation count, which can stay high for years after someone has effectively stopped publishing.

DBLP: the field's own bibliography tool, not just Google Scholar

Most supervisor-search advice points at Google Scholar and stops there, but computer science has its own purpose-built bibliography database that Scholar doesn't replace: DBLP, which indexes roughly 6.7 million publications organized specifically around CS conference proceedings and journals, with conference and workshop papers making up close to half of that total. Searching a professor's name on DBLP surfaces a clean, dated list of exactly what they've published where, without the noise Scholar sometimes includes from unrelated co-authors sharing a name, or from preprints that never went anywhere. It's a five-minute cross-check worth running alongside Scholar and arXiv, not a replacement for either.

arXiv's 2025 crackdown changes what a preprint proves

Since October 2025, arXiv's own CS category has stopped accepting review and position papers unless they've already been peer-reviewed elsewhere, a policy change arXiv introduced after a flood of AI-generated "annotated bibliography" style submissions started clogging the category. Coverage through 2026 has described arXiv escalating its response to AI-generated submissions further since then. The practical upshot for a cold-emailing applicant: an arXiv listing is still a genuinely useful, fast-moving signal that a professor is actively working, but it's worth actually opening the paper rather than just counting recent postings, since the category is now under tighter scrutiny specifically because volume alone stopped being a reliable signal of real research output. The Computing Research Association has long treated peer-reviewed conference papers as the field's equivalent of a journal article for evaluation purposes, which is the underlying reason a professor's conference acceptances, not just their preprint count, are the more durable signal to check.

A practical search order

• Identify professors publishing recently in your specific subfield, cross-checking arXiv and their last one to two years of conference acceptances, not just their overall citation count.
• Check whether the program you're targeting admits directly into a named advisor's group or runs a rotation-style first year, since this changes how specific your application's research statement needs to be.
• Email the professor directly, referencing a specific recent paper of theirs, and ask plainly whether they currently have funding for a new student, since grant cycles mean this genuinely varies year to year.
• Separately check whether portable funding, an NSF GRFP, a UKRI studentship, or a CGRS-D award, applies to your citizenship and target country, since a portable award changes your leverage in the conversation.

GradScoutFunding is built for that first step: it searches professors by field and country, confirms they're actively publishing, and drafts a personalized first email grounded in their real recent paper for you to review and send yourself, it never sends anything automatically. The free tier gives 2 credits with no card required, and paid credit packs are one-time purchases that never expire.

Common questions