Namma AI
A grade-wise CT & AI learning platform for schools
Zinwave is designing and building Namma AI’s online learning portal for the new CBSE Computational Thinking & AI curriculum (Grades 3–8) — a grade-wise platform with modules, activities, dashboards and AI-assisted review, engineered to scale to tens of thousands of students across schools.
The challenge
Namma AI needed a production-grade portal to deliver the new CBSE Computational Thinking & AI curriculum across Grades 3–8: grade-specific learning journeys, teacher and principal dashboards, evidence and reporting — reliable during school hours and built to scale to many schools and 20,000+ students. Because it holds data about children, it also had to be secure and privacy-compliant from day one.
What we did
- Built the activity engine as the architectural centrepiece: 16 config-driven renderers behind a single `ActivityDefinition` contract. A new activity type is a configuration record, not a release, which is what lets the curriculum team author across six grades without engineering in the loop.
- Structured the whole system as a Turborepo + pnpm monorepo — `apps/web`, `apps/api`, a token-driven `ui` package that is the Figma target, the activity contract and engines as separate packages, and shared Zod schemas so a DTO cannot drift between frontend and backend.
- Made the portal a React + Vite PWA, role-routed for students, teachers, coordinators and principals. PWA was a requirement, not a flourish: these are low-end school devices on weak Wi-Fi, and the app has to stay usable when the connection does not.
- Ran the backend as a NestJS modular monolith on AWS Fargate in the Mumbai region, with infrastructure declared in AWS CDK. Region choice is a compliance decision as much as a latency one — children’s data stays in-country under India’s DPDP Act.
- Split content between global and per-tenant with explicit versioning, so a school can be onboarded without cloning curriculum, and curriculum can be revised without rewriting history under students already working through it.
- Combined objective auto-grading with AI-assisted review of written answers, with teacher override always available. The teacher stays the authority; the AI reduces the marking load.
Tech & scope
The outcome
- A single grade-wise platform covering learning, projects, evidence and certificates for Grades 3–8.
- A reusable content engine that lets Namma AI scale from one grade to all six without rebuilding.
- Production-grade foundations — audited, DPDP-compliant and role-based — engineered for 20,000+ students.
Building something like Namma AI?
Tell us what you are trying to ship. You will get an honest review, a clear plan and a fixed, affordable quote, usually within a day.