Job search
AI Fellowships and Residencies vs. Full-Time Research Jobs: How to Build Your Shortlist
Compare time-limited AI fellowships and residencies with ongoing research jobs by employment status, project scope, eligibility, location, and conversion uncertainty.
An AI fellowship, residency, and full-time research job can all involve hands-on research, mentoring, code, experiments, and papers. The label alone does not tell you whether you will be a salaried employee, receive a stipend, remain enrolled in school, or have a job after the program. Compare the written terms: payment and legal status, time commitment, project output, eligibility, location, and what happens when the opportunity ends.
The current examples make the distinction concrete. OpenAI describes its six-month Residency as full-time employment and says residents cannot be actively enrolled in school while employed; its 2026 application is closed. Anthropic’s January Fellows cohort is a four-month full-time funded program, with a posted October 18, 2026 deadline, country-specific benefits, and no guaranteed full-time offer. An ongoing Research Scientist posting describes a broader role rather than a fixed cohort. These are examples from different programs, not universal definitions. You can browse recent research and engineering roles on LandOffer.ai as you build a shortlist.

Start with the opportunity’s structure
Think of the label as the heading on a posting. The body tells you the arrangement. A fellowship may provide a stipend, benefits, compute funding, and mentorship for a defined research project. A residency may be a fixed-length employee role designed to help someone move into a research area. A full-time research scientist job may have no cohort end date and ask you to own a longer-running research agenda. Employers use terms differently, so verify the actual terms every time.
| What to compare | Time-limited fellowship | Employee residency | Ongoing full-time research job |
|---|---|---|---|
| Relationship and payment | Program-specific funding or stipend; confirm the written status, benefits, and payment rules | May be an employee role with salary and employee benefits | Employee compensation and benefits, as stated in the role and offer |
| Time horizon | Fixed cohort, often with an application cycle and end date | Fixed program length; may have rolling start dates | Ongoing role without a pre-set cohort end date |
| Expected work | A scoped empirical project, program deliverable, or public research output | Work alongside teams with structured mentorship and a transition focus | Research responsibilities and agenda defined by the position and team |
| Eligibility constraints | Topic fit, cohort timing, work authorization, and location may be central | Enrollment, availability, location, and transition timing can be decisive | Role-specific skills, seniority, location, and authorization requirements |
| Next step | Completion does not automatically create a permanent job | A later offer may be possible, but check whether it is promised | The job itself is the ongoing employment opportunity |

Use this as a question list, not a ranking. A funded fellowship may offer focused mentorship and room to publish, but it may have strict location, work authorization, or full-time availability terms. A residency can combine salary with a defined transition program, yet its eligibility may conflict with your current degree enrollment. An ongoing job may provide continuity but expect you to arrive ready to carry the responsibilities of the role. The right option depends on which constraint you can meet and which work you want to do next.
A residency can already be a full-time job
OpenAI’s Residency shows why “residency versus full-time job” is not always a clean split. The company describes it as a six-month program for builders and early researchers, with mentorship and research work alongside OpenAI teams. Its official page says residents are full-time employees and receive a full salary. It also says an actively enrolled student cannot participate while employed, and that high-performing residents may receive an offer to join full time after the program. The page does not promise that offer to every resident.
That structure differs from a university course or a short workshop. If you are in a degree program, compare the residency’s enrollment rule with your leave, funding, and return plans before applying. Ask whether you would need to pause enrollment, finish a degree first, or choose another start window. Do not assume that a flexible start date solves the school-status question. OpenAI says its 2026 Residency application is closed, so this page is useful for understanding the program structure, not as evidence of an open current application.
An “employee residency” can still be a temporary job. Read the dates and follow-up language separately: duration, salary, benefits, whether the employer may extend the program, and what consideration for a later role means. “May receive an offer,” “pathway,” and “opportunity to be considered” do not mean an automatic conversion. If continued employment is important to your plan, shortlist permanent positions too rather than treating the residency as a guaranteed bridge.
A fellowship can fund research without promising a job
Anthropic’s current AI Safety & Security Fellows posting describes a four-month, full-time program. It includes direct researcher mentorship, research expenses such as compute funding, and a goal of public output such as a paper submission. The company lists a weekly stipend and says benefits vary by country. The posting says applicants must have work authorization in the United States, United Kingdom, or Canada and be located in one of those countries for the program; it says the program cannot sponsor visas. It also says a full-time offer is not guaranteed.
For the January cohort, the posting says applications are open through October 18, 2026, with rolling review. That date is an example of how a cohort deadline can matter more than a general careers-page alert. Check the live posting before you apply: a deadline, workstream, stipend, eligible location, or portal can change between cohorts. In this case the Greenhouse listing also directs applicants to complete the partner application form; simply submitting the Greenhouse form is not enough. Follow the current instructions on the official page.
Do not infer that stipend and salary are interchangeable. Ask who administers payment, what “benefits” mean in your country, whether the program has a written agreement, and whether you can take other work. Confirm the tax, insurance, academic, and work authorization implications with the relevant institution or qualified professional. The public description can tell you what the organization advertises; it does not answer every participant’s contractual or immigration question.
The research output matters too. A public paper goal can be attractive if you want to demonstrate research execution, receive close feedback, and explore a new field. But a publication goal is not a publication guarantee. Ask whether projects are assigned or chosen, how mentors are matched, what data and compute are available, who owns the work, and how results may be released. Anthropic’s current posting describes distinct workstreams and a project-selection and mentor-matching process. That tells you more about the actual experience than the word “fellow” alone.
Compare the research work, not just the program package
A short program and a permanent role may share the same technical topic while expecting different levels of independence. OpenAI’s Research Scientist posting says applicants may be considered for roles across teams and describes developing or improving machine-learning ideas and owning a research agenda with long-running projects. Google DeepMind describes Research Scientists as formulating hypotheses, designing and evaluating models, and often contributing to foundational papers; it also describes Research Engineers as building and scaling systems to test ideas. These role pages are broad descriptions, so a specific team posting still controls.
Anthropic’s current Research Scientist, Interpretability role gives a more concrete example of ongoing work: designing experiments, developing methods for understanding models, building research infrastructure, and communicating results internally and publicly. It is a role-specific posting in San Francisco with its own qualifications. Its responsibilities are not necessarily the same as a four-month fellows project, even when the research area overlaps.
Use a small work-sample map. For each opportunity, write down one likely deliverable, the level of ownership expected, the feedback cadence, and the audience for the result. Then connect that to evidence you already have. If you built an evaluation harness, point to the benchmark design, failure analysis, and checks you performed. If you ran a research project, identify the hypothesis, method, and what the result does not establish. Avoid telling a permanent-job employer that you want only a short training experience if the position asks for long-term ownership; avoid telling a fellowship program that you expect a full-time job if the program explicitly says conversion is not guaranteed.
Build a shortlist that fits your constraints
Score each opportunity on six fields, keeping the evidence beside your judgment:
- Status and payment: salary, stipend, grant, or another arrangement; who pays; which benefits are specified; and what remains unclear.
- Duration and availability: start and end dates, full-time hours, extension terms, and whether the program conflicts with school, a current job, caregiving, or other commitments.
- Eligibility: degree or experience requirements, enrollment rules, skill expectations, and whether they are minimums or preferences.
- Research output: the project, expected deliverable, publication rules, data and compute access, intellectual-property terms, and mentor structure.
- Location and authorization: where you must live and work, office attendance, travel, and work authorization or sponsorship terms.
- After the end date: whether a new job is guaranteed, merely possible, or not part of the program; and which permanent roles you can apply to directly.
Mark each field “confirmed,” “unclear,” or “conflict.” A program with excellent topical fit may still be a poor choice if its location or full-time requirement does not work. A full-time job may fit your schedule but require a research specialty you do not yet have. Separate a promising research topic from the practical conditions needed to take part.
Here are two fictional shortlist decisions. Nia is a PhD student who wants to remain enrolled and complete her dissertation on schedule. The six-month OpenAI Residency’s no-active-enrollment rule is a direct conflict unless she can make an approved change to her education plan. She should screen for a fellowship or internship whose terms allow her intended enrollment, verify those terms with her university, and keep permanent research jobs on the list for after graduation. She should not treat any “residency” title as an exception to an employer’s stated rule.
Marco is a research engineer who has finished school and can relocate for a fixed-term project. He prefers a defined public research deliverable and can meet the current Anthropic Fellows location and work authorization conditions. He may place the four-month program high on his list, but he should compare its stipend, benefits, and end date with an ongoing Research Engineer or Scientist role. Since the fellowship does not guarantee a full-time offer, he should submit a separate application to a permanent role if its responsibilities match his experience. These fictional examples show a decision method, not real applicants or a prediction of who would be selected.
Check the posting state and next step
Before spending time on materials, open the employer’s current careers page and confirm that the exact opportunity is open. Record the title, workstream, location, date checked, and application deadline. An older announcement can explain the program’s purpose but should not override a current role page. If the role says “expression of interest,” find out whether it is tied to a funded position or simply collects interest for future contact. If the posting names an external application partner, finish the official partner steps and retain a copy of the confirmation.
Ask direct questions when the public page is silent: Is the payment a wage or stipend? Who provides benefits? Can current students remain enrolled? What is the expected weekly commitment? Does the program guarantee a later job? Which immigration support is available for this specific program? What happens to the project’s code, paper, or data after the end date? These questions do not imply a problem; they help you decide whether the terms fit before you commit.
Keep two shortlists if your timeline allows it: time-limited fellowships or residencies with fixed dates, and ongoing research roles with direct applications. Revisit the first list when a cohort opens and the second whenever a matching position appears. For each one, preserve the source URL and the date you checked it, because program details can change quickly. Then browse recent research and engineering roles on LandOffer.ai and compare the live role description against your work, eligibility, and location requirements.



