HyperQuark Research Fellowship (HQ-W27)

**ABOUT THE FELLOWSHIP** HQ-W27 is the second cohort of the HyperQuark Research Fellowship: remote, twelve weeks, from 4 January to 28 March 2027. Its theme is evidence over claims: are the AI systems that read, match, grade and source people accurate, fair and reliable? Each track tests one new claim on public or synthetic data with open models, to the standard of a publishable paper. **RESEARCH TRACKS** Judges You Can Trust: whether AI judges that agree with human graders on standard English still mark down the same answer written in Indian English. Fair Matching, India Edition: whether bias in AI résumé screening runs through caste-coded names or through proxies such as college tier, city tier and English variety. India Skills and AI-Exposure Graph: a validated map from India's occupation codes to skills, and whether India's own task descriptions change which jobs AI touches. Reliable Recruiting Agents: whether invented candidates and unsupported claims explain why sourcing agents that succeed once fail on later runs. **HOW IT RUNS** Twelve weeks in five phases: problem comprehension, direction lock, a quality gate at week 5, defence and validation, then write-up and release. A research pack opens each phase. Every fellow hands in a short report every Sunday, and track leads answer it in writing. It takes about six to eight hours a week, with one weekly cohort session. **WHAT FELLOWS PRODUCE** For each track, a paper (or an honest retrospective), a public release of what it built with the code to rerun every number, and a recorded demonstration, published on the lab's site with the fellows named as authors. **DATA AND ETHICS** Public or synthetic data only, open-weight models and published code No data collected from people, no scraping of people and no experiments on people Never linked to insiderOne member records **FELLOWSHIP AGREEMENT** Every fellow signs the HyperQuark Research Fellowship Agreement before taking part. Under it, all rights in the work done in the fellowship belong to Aurist Private Limited, which runs insiderOne and the lab, and may be used in any product, tool or research without payment or approval. Fellows who make a substantial contribution are named as authors of what is published from their work, and may submit their own papers after the lab's review. Read it before you apply: hyperquarklabs.insiderone.in/fellowship-agreement **WHAT FELLOWS RECEIVE** Weekly written feedback, authorship of the paper and release they contribute to, and a certificate anyone can verify by its ID. The fellowship is unpaid and is not employment. **WHO SHOULD APPLY** We look for curiosity, rigour and the ability to finish. Backgrounds in machine learning, natural language processing, statistics, economics and labour data, fairness and AI policy, or software engineering are all relevant. More: hyperquarklabs.insiderone.in/fellowship