SETU

I spent nearly eight years living full-time in a TGTWREIS Gurukulam — a government-run tribal welfare residential school in Telangana. I watched a headmaster write a letter about a broken water pump. Students waited over a month for it to be resolved. Not because anyone didn't care — but because that letter passed through five offices, with no tracking, no acknowledgement, and no way to know if it had even been read.

That is not a rare case. It is how the system works.

SETU is my attempt to fix it.

When a school reports an issue today, it enters a paper chain that can take weeks or months to reach someone with the authority to act. There is no digital trail. No acknowledgement. No accountability. A water shortage, a damaged dormitory roof, an emergency fund request, a recurring electricity failure every monsoon — all treated the same way, all subject to the same delays, all invisible once filed.

SETU replaces that chain with a structured, trackable, digital reporting system. A local NLP pipeline (spaCy + HuggingFace Transformers) classifies urgency and detects safety-critical language automatically. A weighted priority algorithm — combining issue age, stage stagnation, recurrence history, and urgency — ensures no issue can be deprioritized indefinitely. Every action by every official is immutably logged. Nothing disappears.

The stack is fully open-source — Next.js, Node.js, FastAPI, PostgreSQL, Docker — with no proprietary APIs, built to run in low-connectivity rural environments.

This project is grounded in real field research: interviews with headmasters, wardens, and district officials conducted during March 2026 and committed directly to the repository. Every design decision traces back to a real conversation.