Calibrated peer review that lets colleges grade project work at scale
TL;DR Rubricpeer is grading software for Indian colleges with heavy project, lab and writing coursework. Students review each other's submissions against the teacher's rubric, the system learns which reviewers are reliable, and faculty sign off on every final mark. Colleges pay per student for detailed feedback without extra faculty hours.
Indian higher education is set to grow fast. NEP 2020 aims to lift gross enrolment in higher education to 50% by 2035, from 26.3% in 2018[1]. Faculty numbers won't keep pace, and the work that suffers first is the kind a machine can't mark - design reports, lab write-ups, code reviews and case analyses. In big sections students often get a number and nothing else, weeks after submitting.
I found no public figure for how many assignments an Indian faculty member grades per semester or how long feedback takes. Fill that gap in your own validation interviews before you build anything.
Rubricpeer turns grading into a structured exercise. A teacher uploads a rubric and two or three sample answers she has marked herself. Each student first grades those samples, and the gap between their marks and hers becomes a calibration score. Then each submission goes to four or five classmates, and a statistical model weights every reviewer by past accuracy and bias, the same approach Stanford researchers tested on 63,199 peer grades from a large online-course platform[2]. Students earn part of their grade for reviewing well and the teacher sees only the submissions where reviewers disagree.
I reframed the seed, which proposed paying peers to review and issuing skill badges. Cash rewards invite collusion, and why would an employer trust a badge that students award each other? Grading the quality of each review is the mechanism the peer-review module in a popular open-source LMS already uses[3]. The output is a faculty-moderated portfolio of marked work, not a certificate.
Students use it and the college pays. The buyer is a dean of academics or the head of a department running large project-heavy courses at a private university or autonomous college. The leading Canadian peer-assessment product doesn't publish its prices[4], and I found no Indian price point. My working assumption is ₹300–600 per student per year, which needs testing. A free open-source LMS is the real alternative, so the price has to buy saved faculty time.
NEP 2020 set the 50% enrolment target in July 2020, so enrolment is meant to keep climbing until 2035[1]. AI writing tools also make a single final submission harder to trust, which pushes faculty toward marking drafts and process as well, and that multiplies the load. The DPDP Act 2023 now sets clear consent rules for student data, including guardian consent for any student under 18[5].
There is no public figure for India's spending on peer-assessment or grading software. The nearest signal is abroad - one Canadian peer-assessment product says it serves over 200,000 users across more than 250 institutions[4]. That shows colleges will pay for this category, but it says nothing about Indian budgets.
Treat enrolment growth as context, not as your market size.
A Canadian peer-assessment product sells calibrated reviews with "feedback on feedback" to North American universities and plugs into common learning platforms[4]. The peer-review module in a free open-source LMS already gives students a grade for their reviewing[3], but it needs a patient teacher to set up and has no reviewer-weighting model. Large online-course platforms use peer grading at scale for its online courses[2].
I didn't find an Indian product built for large college sections in my limited checks.
| Difficulty | Moderate (34/100) |
| Starting capital | ₹40–75 L |
| Time to launch | 5–8 months |
The software is manageable. Trust is the hard part, because faculty must believe the weighted marks are fair and students must believe that classmates can't sink them. You'll need a semester of pilot data showing that weighted peer marks land close to faculty marks before any college puts them in internal assessment.
No sector licence applies to assessment software. You need the baseline, plus each university's agreement.
Building this outside India? The categories above usually have a local equivalent, so check with your own regulator.
Don't let AI set final grades. Peer and machine scores are inputs, and a faculty member should own every mark that counts.
Want to work out the AI layer for Rubricpeer? Talk to Yash →
I'd back a founder who has taught in or sold to Indian colleges and can sit through a slow, semester-paced sales cycle. If you're hoping to sell skill certificates to students then skip it. Prove one thing first - across two pilot courses, weighted peer marks fall within one rubric band of faculty marks for at least 80% of submissions.
Plan for roughly ₹40–75 L, mostly on product, pilots with colleges and a long sales cycle.
No sector licence. Each university decides through its own academic bodies whether peer-weighted marks can count, and you need DPDP compliance for student data.
Yes, and calibrated peer-assessment tools already sell abroad. Check local student data rules and how universities there approve grading methods.
The free Rubricpeer validation kit has the line-by-line capital breakdown, the effort scoring, the data-protection checklist, a 14-day validation sprint with pass/fail targets, a 90-day launch plan, 10 interview questions for faculty and academic heads, and 5 assumptions to test. Download the kit →
Thinking about building Rubricpeer or something close to it? The make-or-break bit is getting faculty to trust peer-weighted marks enough to use them, and planning that pilot is the kind of thing I help founders with. If you need a hand, you know where to find me. Talk to Yash →
For illustration only. Names are fictional and ideas hypothetical; sourced facts are real but may be out of date. Nothing here is legal, financial or professional advice, and Yash Khatri is not personally liable for any use of it. Read the full disclaimer.