Science Postdoctoral Fellowship 2027 at the University of Toronto

Last verified: 10 August 2026

Applications for this round close at 5 PM EST on Monday, October 5, 2026. That is eight weeks away, and it is a tighter window than it looks, because you cannot submit anything on your own. A professor at the University of Toronto has to agree to supervise you first, and that professor then submits half of your application package on your behalf.

The award itself is worth the effort. Successful Fellows receive $85,000 CDN per year in salary for two years, plus a further $11,000 CDN per year paid towards benefits and postdoctoral levy costs. It is open to researchers from any country, and prior experience with artificial intelligence is not required.

Start with the supervisor, not the paperwork:

Almost everything else in this application depends on one decision that is not yours alone. You need a University of Toronto supervisor before the deadline, and that person carries real responsibilities in the process.

Here is what the programme requires of them:

• The supervisor must hold either a full time budgetary tenure stream appointment or a status only appointment at the University of Toronto.
• They must be an eligible Principal Investigator under the University of Toronto PI eligibility criteria.
• If your main supervisor does not have methodological expertise in AI, you must add a second supervisor who does.
• A supervisor may back only one applicant as primary supervisor in the whole competition.

Status only professors usually hold their main full time post somewhere else, most often at an affiliated hospital. They still qualify, so do not rule out a researcher simply because their day to day base is not the main campus.

How to find one when you have no contacts at U of T::

The programme is explicit that it does not match candidates with supervisors. That work is yours. A few things that actually help:

• Search the U of T Discover Research portal by research interest rather than by department name. It surfaces people you would never find by browsing faculty pages.
• Read recent departmental pages in Arts and Science and in Engineering, and note who is already publishing work that borders on your topic.
• Ask your PhD supervisor for introductions. A forwarded email from a known colleague opens doors that a cold message will not.
• Write to potential supervisors with a short, concrete project idea attached, not a general request for a position.

One more point that shapes who you should approach. The programme states that it gives priority to applicants proposing a new postdoctoral project with a new supervisor. Staying in the lab where you did your PhD works against you here.

What the fellowship pays, and what it does not:

Read this section closely, because the funding has clear edges and applicants regularly assume it covers more than it does.

What you receive:

• $85,000 CDN per year in salary.
• A further fixed $11,000 CDN per year towards the standard benefit rate and the postdoctoral levy costs incurred by your supervisor or the awarded unit.
• A term of two years.
• Modest additional funding during your tenure for conference travel and research training activities.
• Structured training in AI and related computational skills.

What it does not cover:

• Research costs. Consumables, field work travel and similar expenses are not paid by the Schmidt AI in Science training programme. Your supervisors are responsible for those.
• Any salary or benefits costs above the supplemental $11,000 CDN per year. That gap falls to the host supervisor to fund.

This is worth raising with your prospective supervisor early and directly. A lab that cannot cover your consumables is not a lab that can host your project, however enthusiastic the reply to your first email was.

Who can apply:

The eligibility rules are strict on dates and disciplines, and generous on nationality and AI background.

Your PhD completion date

• You must have completed all requirements for your doctoral degree no earlier than January 1, 2024.
• If your career was significantly interrupted within two years of finishing, for parental, medical or family related reasons, that window extends back to January 2023. If this applies to you, add a short section at the end of your candidate statement explaining the interruption.
• International candidates must have completed the PhD before the fellowship start date.
• Canadian citizens and permanent residents must complete it within six months of the start date.
• Completion means finishing every requirement of the degree. It does not mean convocation. If you did not have proof in hand when you applied, you must supply it within three months of completion.

Your field of study

Your degree must sit in a natural sciences or engineering discipline, which includes computer science and mathematics. If you are unsure whether your field counts, check the NSERC Research Subject Codes, sections 1000 through 7000.

You do not need an AI background. What you do need is domain expertise, a strong record, and a credible desire to learn AI methods that could speed up discovery in your field.

Who is ruled out

• Anyone holding another postdoctoral fellowship at the same time. If you hold one when the award is made, you must resign from it before taking this up.
• Anyone who holds or has held another prestigious named U of T postdoctoral fellowship. This includes competitive postdoc programmes run by the Acceleration Consortium, the Faculty of Arts and Science, the University of Toronto Mississauga, the University of Toronto Scarborough, the Data Sciences Institute, and the Provost’s Office.
• Anyone who has held a position as a Schmidt Science Fellow at the University of Toronto.

Projects that will not be funded, no matter how strong they are:

This is the section that saves people the most wasted effort, and it is buried deep in the official FAQ where most applicants never reach it.

The programme funds research that uses AI methods to answer domain questions in the natural sciences and engineering. Two categories of project fall outside that, regardless of quality:

• Health and medical science projects. If the primary objective is to improve health, produce better health services or products, or strengthen the Canadian health care system, the project is not eligible. The only route in is if the primary objectives genuinely advance knowledge in a natural science or engineering discipline instead.
• Projects that build AI itself. Computer science and engineering projects are welcome, but only where AI methods are being used to advance fundamental knowledge in the domain. If developing new AI tools and technologies is the primary research objective, the project is not eligible. The same applies to engineering projects focused mainly on building AI optimised tools, applications and technologies.

The distinction is about direction of travel. AI as the instrument is fine. AI as the destination is not.

The application package, piece by piece:

Everything goes by email to the Schmidt Sciences Program Manager at schmidtfutures@utoronto.ca, with this exact subject line for both you and your supervisor:

APPLICANT NAME: SCHMIDT FELLOWS PDF APPLICATION

Incomplete applications are not accepted. There is no grace period and no chance to send a missing file afterwards.

What you submit

Sections a through g go in as a single PDF file:

• a. The Schmidt AI in Science Postdoctoral Fellowship Application Form. Instructions for the narrative sections sit inside the form.
• b. Research proposal, 3 pages maximum.
• c. Bibliography, 1 page maximum.
• d. AI in Science training plan, 1 page maximum.
• e. Candidate statement, 2 pages maximum.
• f. Significance of leadership contributions, 1 page maximum.
• g. Your CV.

Separately, you must complete the Applicant Demographic Survey. Answers stay confidential and play no part in assessing your application, now or in future rounds.

The reference letters, and the date nobody notices

You need two confidential letters of reference, each 2 pages maximum, covering your research excellence, the merit of the proposed research, and the suitability of the proposed research and training environment. Referees must state their relationship to you in the letter.

The mechanics here trip people up:

• Letters do not go to the programme. They go directly to your proposed supervisor.
• Your supervisor then includes them in their own submission package on October 5.
• The programme asks that referees send them no later than September 28. That is your real deadline for references, a full week before the public one.
• Your proposed supervisor cannot be one of your referees.

Identify and confirm your referees early, and give them the submission instructions in writing. Chasing an academic for a letter in the last week of September is a losing game.

What your supervisor submits

• The Supervisor’s Assessment Form, completed by the primary supervisor, together with a statement of endorsement of up to 2 pages. Instructions are nested inside the supervisor form.
• Your two reference letters, collected on your behalf.

Also Check: Professional Fellows Exchange Program for Governance and Society 2026 in USA (Fully Funded)

How your application is actually scored:

The selection committee draws on experts from across the natural sciences and engineering at U of T, and the committee chair also acts as Equity Officer. Four weighted criteria decide the outcome.

Quality of the research, 35 percent. Assessed from your research proposal.

• Clarity and significance of the research question and the scientific gap it addresses
• Scientific rigour, innovation and feasibility, showing high potential for success
• Depth and appropriateness of how AI and machine learning methods integrate with the scientific domain
• Potential for broader impact, scalability and contribution to the field

Strength of the candidate, 35 percent. Assessed from your candidate statement, leadership contributions, CV and reference letters.

• Research excellence and potential, judged relative to your career stage: publications, awards, recognition, citations and trajectory
• Clarity of motivation and career vision, and evidence the fellowship advances your long term independence
• Leadership and broader contributions, including mentorship, outreach and community building beyond research
• Commitment to equity, diversity and inclusion

Synergy between applicant, supervisors and institutional environment, 20 percent. Assessed from the supervisor statement, your candidate statement and your proposal.

• Demonstrated commitment and capacity of your supervisors to actively support you
• Quality and richness of the mentorship environment and available resources
• Your supervisors’ expertise and track record with AI and machine learning tools
• Alignment between your background, their expertise, the institutional environment and the proposed research

AI adoption and training plan, 10 percent. Assessed from your methodology, training plan and supervisor statement.

• An honest assessment of your current AI knowledge and your capacity to develop what the project needs
• Clear articulation of the AI skills you will build during the fellowship
• Training activities that are appropriate and ambitious relative to your goals
• Evidence that your supervisors and the institution can actually deliver that training

The scoring bands

Each application is scored on a five point scale, and there is a hard cut off.

• Outstanding, 4.5 to 5.0: excels in most or all aspects, shortcomings minimal, innovative, fills a critical gap, and the investigator is well positioned for success.
• Excellent, 4.0 to 4.4: excels in many aspects and reasonably addresses the rest, very interesting, makes important advances, minor limitations noted.
• Good, 3.5 to 3.9: excels in some relevant aspects and reasonably addresses most others, has strengths, but moderate limitations and some improvements are necessary.
• Fair, 3.0 to 3.4: broadly addresses the relevant aspects, has merits but many limitations, major revisions required. Will not be funded.
• Poor, 0.0 to 2.9: fails to provide convincing information, or has serious inherent flaws or gaps. Will not be funded.

Anything below 3.5 is out of contention. In practice, a merely competent application does not lose narrowly here, it fails a threshold.

Also Check: Commonwealth Fellowships 2027 in the UK for Commonwealth Citizens

Where applications quietly lose marks:

Nothing in this section is a formal rule. It is what the weightings tell you if you read them as instructions rather than as description.

• The training plan is only 10 percent, and it is the easiest 10 percent to leave on the table. It is one page. Name specific methods, specific courses or workshops, and specific people who will teach you. Vague statements about becoming familiar with machine learning score poorly against a criterion that explicitly asks for articulated skills and competencies.
• Synergy is worth 20 percent and it is not written by you alone. Your supervisor’s endorsement and your own statement have to describe the same project, the same fit and the same reasons. Two documents that read like they were written by strangers cost you a fifth of the score.
• A single supervisor needs a justification. The programme prioritises pairs, one domain expert and one AI expert. If you are proposing to work with just one person, they must justify that in the supervisor form and describe their relevant AI expertise. Do not leave that box for your supervisor to discover on October 4.
• Leadership is scored separately from research. The dedicated one page section covers mentorship, outreach and community building. Applicants with strong publication records often treat this as filler, then lose points inside the 35 percent candidate criterion.
• Equity, diversity and inclusion is a listed criterion, not a courtesy. Write about what you have done, not about what you believe.

Using generative AI in your application

New for this cycle, the programme published explicit guidance, and it is more permissive than many applicants expect.

You may use generative AI tools to prepare your materials, whether to work more efficiently, to support writing in a second language, or to streamline drafting. This mirrors the position taken by the Canadian granting agencies.

The conditions matter, though:

• You remain fully responsible for everything you submit.
• Every claim must be true, accurate and complete.
• All sources must be properly acknowledged and referenced.

Applicants are also directed to the University of Toronto’s own AI guidelines and to its Policy on Ethical Conduct of Research Involving Human Subjects for wider research integrity and ethics expectations. Treat AI as a drafting aid you then verify line by line, not as an author.

Key dates for this cycle:

• September 28, 2026: target date for referees to send letters to your proposed supervisor
• Monday, October 5, 2026, 5 PM EST: application deadline for both applicant and supervisor packages
• November 2026: selection committee adjudication meeting
• January 2027: notifications sent to applicants
• May 1, 2027 to January 1, 2028: window in which the fellowship must begin

Frequently asked questions:

This is a well paid, two year, fully salaried postdoctoral position at one of Canada’s strongest research universities, open to any nationality, and deliberately designed for domain scientists who want to learn AI rather than for people who already build it. Those conditions make it unusual, and competitive.

Your first move is not writing. It is email. Identify two or three realistic supervisors at U of T this week, check whether each has AI expertise or would need a partner who does, and find out whether they have already committed to another applicant, because they are allowed only one. Everything else in this application can be written in eight weeks. A supervisor cannot be conjured in the last one.

Full details and the official forms are on the University of Toronto Schmidt Fellows application page. Always confirm dates and figures there before you submit, since programme terms can change between cycles.