Ranks Bhopal’s seven ghats against CPCB Class B bathing limits and says whether each fails before, during or a week after immersion.
The problem
Every Ganesh and Durga festival, the MP Pollution Control Board samples the water at Bhopal's seven immersion ghats before, during and one week after immersion, then files the results in 90 to 120 page PDFs that is rarely compared against the legal limits or across years. The BMC officer who plans Durga visarjan on October 2026 must decide where to send cleaning crews, aeration and extra testing.
How it works
Ghat Pehredaar (घाट पहरेदार) turns three years of MPPCB readings into a ranked list of the seven ghats, worst first, scored against CPCB Class B bathing limits. For each ghat it answers three questions, each tied to one lever the officer controls: already failing before immersion (aerate first, or do not use), clearly worse during immersion (cranes, liners, crowd control), still failing one week after (cleanup and re-test). Every number links to the MPPCB report page it was printed on. Rank uncertainty is shown from 1,000 resamples of the six festivals. Missing data is shown as missing, never as clean, including the facts that total coliform was never tested and that 2025 metal readings print the lab's detection limit as if measured. The officer gets a plain-language reason per ghat and a one-page brief to forward on WhatsApp, in Hindi and English.
How Claude was used
Claude Code built the project's core scoring, deployment, the bilingual UI and the Azure deployment.
Claude writes the officer brief from the computed evidence table in Hindi and English with a report page cited on every reading, and a validator rejects any number in the brief that is not present in the data, so the brief can never carry an invented figure.