HyperQuark Research Fellowship
A remote, unpaid research fellowship for students and researchers in eligible countries. Fellows work in tracks on open questions about AI and people, report every Sunday, defend their work, and are credited by name on what the lab publishes.
- Format
- Remote
- Pay
- Unpaid, educational
- Length
- 12 weeks
- Time
- 6 to 8 h a week
- Certificate
- Verifiable by ID
HQ-W27 · Applications open until 20 Dec 2026
Apply to be part of HQ-W27
Evidence over claims: are the AI systems that read, match, grade and source people accurate, fair and reliable? · Starts 4 Jan 2027 · 12 weeks · Remote · Unpaid
An unpaid, educational fellowship for students and researchers in eligible countries. Fellows sign the Fellowship Agreement before taking part. Read it before you apply.
Philosophy
How a fellow works
- 01LearnRead the field before touching it.
- 02ExploreMap the systems, rules and prior work.
- 03QuestionState one question in one sentence.
- 04BuildMake the smallest thing that can answer it.
- 05TestMeasure it against a baseline.
- 06ChallengeDefend it to people trying to break it.
- 07ReviseFix what the challenge found.
- 08CommunicateWrite it up, including what failed.
- 09ContributePut your name to what the lab publishes.
How a cohort runs
One question per track, five phases
- Research packEach phase opens with readings, questions, what to build and what to write.
- Sunday reportEvery fellow hands in a short report by 23:59 Anywhere on Earth.
- Written reviewTrack leads and mentors read every report and answer it in writing.
- Review BoardEach track's final work goes to the HyperQuark Review Board, an internal review, before release.
What fellows get
What you take away
- Your name on the research the lab publishes from your work.
- Credit by contribution, recorded with the CRediT roles.
- Written feedback every week, and a review of your final work.
- A certificate anyone can verify by its ID.
What it is not
Stated plainly
- Not employment.
- Not a route to a job at insiderOne.
- Not access to insiderOne members' data.
- Not ownership of the work: under the Fellowship Agreement it belongs to the company.
- Not an academic qualification.
Cohorts
Past and upcoming
Questions
Before you apply
Is the HyperQuark Research Fellowship paid?
No. It is an unpaid, educational research fellowship, not employment, and it pays nothing now or later. Fellows receive mentorship and written feedback, access to the lab's studio and guidance on free compute for the work, a certificate anyone can verify, and credit by name on what the lab publishes from their work.
How much time does it take?
About six to eight hours a week for twelve weeks, at your own schedule, with one weekly cohort session and a short report each Sunday. It is built to fit alongside study.
Where is it based?
It is fully remote. Fellows take part from home, in an eligible country; the lab's work is coordinated from the San Francisco Bay Area.
When is the next cohort, and when do applications close?
The next cohort, HQ-W27, runs from 4 January to 28 March 2027, with about 12 fellows. Applications close on 20 December 2026. Apply through insiderOne from the cohort page on this site.
Who can apply?
Enrolled students and unaffiliated researchers (not employed to do research in the track's field), aged 18 or over, who live in and take part from an eligible country: India, Singapore and Japan (students and unaffiliated researchers); South Korea (enrolled students aged 19 or over); the United States (enrolled students only, not resident in California or New York); and the United Kingdom, Canada and Australia only where the fellow's course requires the placement and their institution confirms it in writing. Germany, France, the rest of the European Union and the European Economic Area, Israel and every country not named here are not eligible for HQ-W27. No PhD is needed: we look for curiosity, rigour and the ability to finish, and backgrounds in machine learning, natural language processing, statistics, economics and labour data, fairness and AI policy, or software engineering are all relevant.
What will HQ-W27 fellows research?
Four tracks, each testing one new claim on public or synthetic data with open-weight models: whether AI judges are fair to Indian English, whether bias in résumé screening runs through names or through proxies such as college tier, a validated map from India's occupation codes to skills and AI exposure, and why AI sourcing agents that succeed once fail on later runs. Each track works towards a paper and a release of its data and code, for the lab to publish.
What data do fellows use?
Only public, open-licence and synthetic data, with open-weight models. Fellows never receive insiderOne product or user data, in any form, and collect no data from people.
Do fellows get credit for their work?
Yes. The lab decides whether, what and when anything from the fellowship is published. On what it publishes, every fellow who made a substantial contribution is credited by name as an author, and others who helped are acknowledged by name.
Who owns the work?
Aurist Private Limited, which runs insiderOne and the lab. Every fellow signs the Fellowship Agreement (HQ-FA 2.0) before taking part: everything made in the fellowship belongs to the company, which may use it in any product, tool or research, anywhere, without payment or approval, and which alone decides whether, what and when anything is published. Fellows are credited by name on what it publishes from their work and keep what they owned before they joined. Taking part gives no claim, now or later, to payment, equity, a job or ownership. The agreement is published on this site; read it before you apply.
Is weekly work scored by AI?
An AI model drafts a score for each weekly report and the week 5 gate, against the published rubric. The programme director reads every draft, changes it where needed and releases it; nothing a model drafts reaches a fellow before that. Final papers are scored by the people on the HyperQuark Review Board.
What does a fellow receive at the end?
A certificate with a public ID that anyone can check on this site, for fellows who meet its requirements, and credit by name on the paper and release the lab publishes from their work.
Anything else: contactus@insiderone.in
HQ-W27 · Applications open until 20 Dec 2026
Apply to be part of HQ-W27
Evidence over claims: are the AI systems that read, match, grade and source people accurate, fair and reliable? · Starts 4 Jan 2027 · 12 weeks · Remote · Unpaid
An unpaid, educational fellowship for students and researchers in eligible countries. Fellows sign the Fellowship Agreement before taking part. Read it before you apply.