Perplexity just restructured its flagship AI research program into two paid tracks worth up to $220,000 a year, and the priority application window closes on September 30. That is about a week away. If this fellowship has been sitting in your bookmarks tab, this is the deadline that actually matters.
What the Perplexity Research Fellowship Actually Is
The Perplexity Research Fellowship is a paid, full-time program built for early career researchers, engineers, and analysts from any technical or quantitative discipline, not just people who already have an AI or machine learning background. Perplexity says it explicitly: physicists, cognitive scientists, quantitative finance professionals, and theoretical mathematicians are all fair game, as long as you bring a rigorous, analytical mind and the willingness to learn AI research fast.
The actual work spans multi-agent planning, continual learning, pre-training, and human-AI interaction, among other research areas. Fellows embed directly inside Perplexity’s R&D teams in San Francisco, Palo Alto, or New York City. This is not a remote program, and Perplexity is careful to clarify it is not an internship either. Fellows are treated as short-term employees for the length of the program, with a real shot at converting to a permanent, research-focused role if you perform well.
What Fellows Actually Get Once They Are In
Perplexity breaks down the experience into three pieces, and they are worth taking at face value since they shape what your three months actually look like.
The research itself is meant to be the whole job. Perplexity says fellows work on hard, critical problems tied directly to the company’s mission, with no side projects pulling focus. Every hour is supposed to go toward research that matters, not busywork or internal tooling.
Mentorship comes from embedding inside existing R&D teams rather than sitting in a separate fellows-only track. You work alongside senior researchers and engineers day to day, and you also join a smaller community of fellows across the Bay Area and New York offices for technical discussions outside your immediate team.
On resources, Perplexity points to state-of-the-art compute and access to some of its largest datasets, which is a meaningful draw if your current setup involves fighting for GPU time. The company also frames publishing as part of the job description, not an afterthought. Fellows are expected to contribute to publications and research artifacts meant to benefit the wider AI research field, not just Perplexity’s own roadmap.
The Two Tracks, and What They Pay
Perplexity splits the fellowship into two tracks based on where you are in your career, not what project you end up working on.
- Track A, Predoctoral Research Fellow: built for exceptional early career talent in a technical or quantitative discipline. Pays $180,000 annualized base salary.
- Track B, Postdoctoral Research Fellow: built for researchers who have recently completed a PhD in a technical or quantitative discipline. Pays $220,000 annualized base salary.
Both tracks run three months, full time, with the option to extend or convert into a permanent role for people who stand out. Perplexity is prioritizing applications for its November 2026 and January 2027 cohorts. The company also offers immigration and visa sponsorship support, though it describes this as dependent on individual circumstances rather than guaranteed for every applicant.
This Used to Be a Single $220K Residency. Here Is What Changed
Here is the part most coverage of this program has missed entirely. Back in June 2026, this same program launched under a different name, the Perplexity Research Residency. It had one flat rate for everyone: $220,000 a year, regardless of whether you had finished a PhD. There was no predoctoral or postdoctoral split. One track, one number.
Sometime between June and now, Perplexity quietly rebuilt the whole thing. The program got renamed from Residency to Fellowship, split into two tracks, and picked up a $40,000 gap between the predoctoral and postdoctoral tiers. Postdocs still land at the original $220,000 rate. Early career applicants without a finished PhD now enter at a lower number than what the program advertised to everyone just a few months ago.
None of this makes Perplexity the villain here. Tiering pay by career stage is standard practice across this entire industry. But if you were eyeing this opportunity back when it was the Residency and assumed the figure you saw still applies to you, check which track you actually land in before you get your hopes up about the number.
Who Actually Qualifies
Perplexity is not asking for prior AI research experience. What it is actually looking for:
- A strong mathematical foundation and a track record of solving hard problems in your own field
- Deep technical expertise in a discipline outside mainstream AI research. Physics, cognitive science, quantitative finance, and theoretical mathematics are the examples Perplexity names directly
- Genuine curiosity about applying that expertise to a brand new domain
- The ability to explain complicated ideas clearly and collaborate with people well outside your specialty
- Demonstrated technical excellence, even if that excellence has nothing to do with AI yet
If your background reads more like a physics thesis or a quant trading desk than a machine learning conference paper, that is exactly the profile Perplexity says it is hunting for.
How to Apply, Step by Step
- Confirm which track fits you. No completed PhD yet in a technical or quantitative field? You are Track A, Predoctoral. Recently finished one? You are Track B, Postdoctoral.
- Check the location requirement before you invest time. This program is fully in-person in San Francisco, Palo Alto, or New York City. If relocating within the next couple of months is not realistic for you, confirm that with Perplexity before you go further.
- Apply through the official portal only. Submissions go through Perplexity’s Ashby application page, not a general careers inbox.
- Let your technical work do the talking. Perplexity’s own instructions say to follow the application steps carefully and let your technical excellence shine. Skip the generic cover letter tone and lead with specific, verifiable work you have actually done.
- Submit before September 30, 2026, for priority review. Applications stay open on a rolling basis after that date, but early submissions get first consideration for the November 2026 and January 2027 start dates.
- Expect an exploratory conversation before anything formal. If Perplexity sees a strong fit, the next step is a conversation about your background and interests, not an immediate technical interview.
- Do not wait around for a status update email. Perplexity’s own FAQ is blunt about this: no progress updates are sent. Add perplexity.ai to your email allowlist so an interview invite does not land in spam.
How It Stacks Up Against OpenAI, Anthropic, and MATS
Perplexity is not the only place throwing six figure pay at early career AI talent right now. Here is how the fellowship compares to the other programs currently open or recently closed in this space.
| Program | Duration | Compensation | Location |
|---|---|---|---|
| Perplexity Research Fellowship | 3 months, extendable | $180K predoctoral or $220K postdoctoral, annualized | San Francisco, Palo Alto, or NYC. In-person only |
| OpenAI Residency | 6 months | Roughly $219,600 a year ($18,300 a month) plus equity | San Francisco. In-person |
| Anthropic Fellows Program | 4 months | $3,850 a week (roughly $200K annualized) plus around $15K a month in compute funding | Berkeley or London. Remote allowed within the US, UK, or Canada |
| MATS Residency | 6 to 24 months | $155K to $285K a year | London, Berkeley, or Washington DC. Hybrid |
A few things jump out here. Perplexity’s fellowship is the shortest of the four at three months, which leaves the least runway to actually finish a meaningful research output before the clock runs out. It is also the only one of the four with zero remote flexibility built in. On the upside, its postdoctoral pay sits right alongside OpenAI’s residency, and its predoctoral track out-earns most PhD stipends most applicants will ever see.
Honest Limitations
Perplexity’s public page for this program stays thin on details that actually matter for a real decision. It does not name mentors, disclose team size, or point to specific research papers past fellows have produced, information Anthropic’s equivalent program publishes openly, right down to author names and project titles like subliminal learning and agentic misalignment. That is a real gap if you are trying to judge the quality of mentorship you would actually get before you uproot your life for it.
Three months is also short by industry standards. Set that against Anthropic’s four months or OpenAI’s six, and Perplexity is asking fellows to produce meaningful research in roughly half the time some competitors allow. Extension exists, but it is discretionary, and Perplexity does not publish conversion numbers the way Anthropic does, which states that over 40 percent of its fellows have gone on to full-time roles.
Visa sponsorship is described as available “based on individual circumstances,” language vague enough that it is worth confirming directly with Perplexity’s recruiting team rather than assuming it applies to your specific case.
The in-person requirement is also worth weighing seriously. Every competing program on this list except OpenAI’s offers at least some remote flexibility. Perplexity’s fellowship offers none, which narrows the realistic applicant pool to people who can relocate to one of three expensive US cities on short notice, or who already live there.
If you know someone with a physics PhD, a quant background, or a cognitive science degree who has been eyeing a move into AI research, send them this before September 30.
Sources
- Perplexity: Perplexity Research Fellowship (official program page)
- Opportunity Desk: Perplexity Research Residency Program 2026 (Jude Ogar, June 17, 2026)
- OpenAI Residency 2026 program listing
- Anthropic: Anthropic Fellows Program for AI Safety Research, applications for May and July 2026 cohorts
- MATS Program: MATS Residency, compensation and program details
Leave a comment