This is a submission for the Hacktoberfest Open-Source AI Challenge Week 2: Side Hustle
What I Built
F.E.R.P.A. Grader is a subscription-ready, stateless micro-SaaS designed for private high school tutors, essay coaches, and freelance educators.
Grading essays is exhausting, but teachers face a massive legal roadblock: strict educational privacy laws (like FERPA) prohibit them from pasting student text into commercial AI tools like ChatGPT, where user data is logged and harvested for model training.
F.E.R.P.A. Grader solves this by offering instant, rubric-based essay critiques using a zero-database, in-memory architecture. The text exists for 3 seconds during inference and is instantly purged.
What my first user said:
"I tested it with a sample student paragraph. Having instant rubric breakdowns without worrying about data retention policies or school compliance makes this something I would actually pay a monthly fee for as an independent tutor."
Demo & Architecture
- Live Application: https://ferpa-grader.vercel.app/
- Frontend: Built with Next.js and Tailwind CSS for a clean, distraction-free educator dashboard, hosted on Vercel.
- Backend: A decoupled Node.js & Express server running entirely in-memory with no database and no user authentication required.
-
AI Core: Routes through OpenRouter to tap into open-weight models (
nvidia/nemotron-3.5-super-vl:free), enforcing strict JSON schemas for structured grades and commentary.
Code
The complete source code is public and open-source under the MIT License:
Jayesh-JainX
/
FERPA-Grader
Privacy-first, instant essay feedback for educators—built entirely without data storage.
🛡️ F.E.R.P.A. Grader (StatelessGrader)
Privacy-first, instant essay feedback for educators—built entirely without data storage.
📸 Demo & Application Preview
Here is a look at the F.E.R.P.A. Grader interface in action. The split-screen dashboard allows educators to input custom rubrics and student essays on the left, receiving instant, structured critiques on the right with a verified "Memory Cleared" privacy indicator.
(If the image doesn't load directly, you can check out the preview assets inside the frontend/public/ directory).
🍎 For Teachers & Educators (The "Non-Tech" Story)
Grading essays is one of the most time-consuming parts of teaching. While artificial intelligence can help speed up the process by offering immediate feedback and rubric breakdowns, teachers face a major roadblock: student privacy.
Under strict educational privacy laws (like FERPA), educators cannot legally paste sensitive student essays into standard commercial AI tools. Those tools store your text on massive corporate servers and often use…
Why Does Open Innovation Matter?
You cannot build a legally compliant educational app using closed commercial APIs.
- True Data Sovereignty: Closed frontier models retain prompt history and train on user input. By routing through OpenRouter to open-weight models, we retain full transparency over how student data is handled.
- Stateless Compliance: Because the code relies on open weights, we can deploy a 100% stateless backend that makes zero database calls—guaranteeing that minors' academic data is never stored or monetized by big tech.
- Sustainable Cost Structures: Independent tutors operate on thin margins. Open-weight models keep API expenses fractions of a cent per essay, making a viable business model possible.
