Ranius: Research Collaboration and Pro Bono Roles

Ranius · Research collaboration and pro bono

Help test whether teachers can be trained online, with AI, in about two months

Ranius is NIJIN’s project to train teachers online with AI. We are looking for research collaborators and for professionals who can work with us pro bono.

Apply on the Ranius site

What Ranius is

Ranius describes itself as education infrastructure that works alongside schools. Its teacher training follows three steps: seeing, shadowing, and practice with AI coaching. The aim is to prepare a teacher in about two months, without requiring a teaching licence. Ranius was founded by Tatsuro Hoshino and belongs to NIJIN Inc.

The problem it works on is global. The Ranius site cites UNESCO: 273 million children are out of school, and the world needs 44 million more teachers by 2030.

460children enrolled in the online school

1,200+teachers trained

~1,000lesson recordings, with consent

Dec 2026planned start of the trial in the US

Figures are those published on the Ranius website. Ranius has been accepted to GSAP, the overseas expansion programme of JETRO and Japan’s Cabinet Office, with StartX as accelerator partner.

Research collaboration

In December 2026 Ranius plans to open an online Japanese-language school in the United States and run a non-inferiority trial there. One group of teachers is trained through the AI process without a licence. The other group is licensed teachers with human mentors. Sample size, unit of allocation, the non-inferiority margin and the outcome definitions are not fixed yet. We want to set them with collaborators.

Study design

Design the non-inferiority trial, starting with what counts as having become a teacher.

Rubrics and reliability

Rubrics for teachers’ lessons and for children’s non-cognitive skills, and how to reach and keep inter-rater agreement.

Training procedures

Seeing, shadowing and practice: how a teacher learns to observe an individual child.

Engineering

Transcription, extraction by rubric dimension, feedback generation, rater-drift detection, and how far automated scoring can be trusted.

The terms are open, with one exception. Authorship, data access, cost sharing, duration and publication are agreed in writing at the first meeting. The one fixed condition is that unfavourable results are published too. Pre-registration and research ethics approval have not been completed yet, and there is no formal agreement with any university so far.

Pro bono and other ways to take part

You can join by approval as a community member, or take one of the open roles. Each role can be done pro bono, part-time or full-time.

Way to take part	Time
Pro bono	2 to 5 hours a week
Contract or part-time	6 to 10 hours a week
Full-time	You can switch to it later

Open roles listed on the Ranius site:

Software Engineer. Video pipeline, rubric scoring, mentor tools.
Research Lead. Trial design and university collaborations.
Assessment and Measurement Specialist. Rubric building and inter-rater reliability.
Business Lead, US launch. The December 2026 school launch in the United States.

What we promise about children’s data

We do not rank children.
No data is used beyond the scope of consent.
Sharing with schools, universities or employers needs the consent of the child and the parent.

How to apply

Fill in the form on the Ranius site. No CV is required. You choose the role and how much time you can give.
30-minute call with Tatsuro Hoshino to confirm the fit.
Access. After approval you join the Ranius space in Teacher Studio.

Work on this with us

If you study teacher education, measurement or child development, or you build AI systems, tell us what you would like to work on.