r0elcompute/sauti-ya-mwananchi

AI-powered civic grievance platform for Mombasa County, Kenya. Citizens submit voice/text/photo reports via Gemma-powered triage — auto-categorized, ward-tagged, and urgency-scored — feeding a leader dashboard for transparent NG-CDF budget allocation. # Sauti Ya Mwananchi **Sauti Ya Mwananchi: Empowering Communities, Driving Real Action.** Sauti Ya Mwananchi is a civic grievance and AI-driven constituency planning platform built for Kenya's NG-CDF (National Government Constituencies Development Fund) framework. It bridges the gap between citizens and their leadership (MPs/MCAs) by turning unstructured community complaints — voice, text, and photos — into structured, urgency-ranked, and actionable governance data. Built with React, Tailwind CSS, Lucide Icons, and Shadcn UI, with a dark-mode glassmorphism aesthetic, role-based auth routing, in-browser audio recording, and a visual Ward heatmap for leaders. --- ## ✨ Key Features - **AI-Powered Triage (Gemma 4):** Automated transcription of Swahili/English/Sheng voice notes, urgency scoring (1–10), and ward extraction from raw citizen submissions. - **Ward Clustering:** Groups individual grievances into actionable, fundable community projects. - **Dual Portals:** Separate, role-enforced experiences for citizens (submitting and tracking grievances) and leaders (MP/MCA/Ward Administrators reviewing and acting on them). - **Multimodal Submission:** Voice recording, photo/video evidence upload, and text description — all tied to a ward and issued a public tracking ticket. - **Executive Reporting:** One-click AI-generated NG-CDF briefs for parliamentary committee review. --- ## 🏛️ Pages & Architecture ### 1. Landing Page (`/`) - Nav bar with links to Problem, How It Works, and Ward Analytics, plus Citizen Access / Leader Portal entry points. - Hero section headlining the platform's mission, with an interactive visual anchor showing live ward status chips (e.g. *"Bamburi • Water Crisis • Urgency 9/10"*). - Problem statement grid: unstructured complaints, lack of hotspot data, unfair CDF budget allocation. - 4-step "How It Works": Citizen Voices → Gemma 4 AI Analysis → Ward Clustering → Leader Action. - Live community impact metrics (voices submitted, active wards, …