MAATI

Inspiration We saw that children in rural areas, especially in Jharkhand, struggle with early education because of two main reasons: the language barrier and the lack of internet connectivity. We wanted to build a solution that teaches them in their mother tongue without relying on expensive hardware or internet access. Our goal was to create an inclusive, stress-free learning environment for primary school students.

What it does MAATI (Multilingual AI-Assisted Teaching Interface) is a 100% offline Progressive Web App (PWA). It teaches Foundational Literacy and Numeracy (FLN) to kids in Class 1 to 5.

Multilingual Support: It supports Santhali (Ol Chiki), Hindi, and English with real-time translation.

Gaju Buddy AI: An on-device AI assistant that gives visual hints instead of direct answers.

Bagiya Rewards: A gamified system that rewards kids with virtual seeds and butterflies, completely removing harsh "red error" screens to ensure a no-fail learning zone.

MannMitti Wellness: Includes simple breathing exercises to check the child's mood before studying.

How we built it We engineered MAATI as a zero-dependency Single Page Application (SPA).

Frontend: Built with HTML5, Vanilla JavaScript (ES6+), and CSS3 to ensure it is extremely lightweight.

Voice & AI: We used the native browser Web Speech API for voice synthesis and recognition, combined with a fast on-device intent classifier.

Storage: All student progress and rewards are saved locally using the browser's LocalStorage. It requires zero server round-trips or API calls.

Challenges we ran into Hardware Constraints: Government school tablets often have very low RAM and crash when running heavy modern frameworks. We solved this by using pure Vanilla JS to eliminate overhead.

Language Limitations: There was a lack of robust native Text-to-Speech engines for the Santhali dialect. We mitigated this by integrating custom phonetic mapping and recorded voice chips.

What we learned We learned that building technology for rural education is not just about writing code; it is about empathy. We learned how to optimize web applications for extremely low-end devices and zero-bandwidth environments. We also gained deep insights into child psychology by designing a "Zero-Password Visual Login" and focusing on positive reinforcement rather than penalty-based grading.