Uchaguzi Safe - Crowd-Sourced Election Monitoring & AI Heatmap Platform

Background: In Kenya's highly competitive democratic landscape, election transparency and civic participation are vital for maintaining national peace, institutional accountability, and public trust. During election cycles, rumors, unverified polling anomalies, and localized security incidents can spread exponentially on social media, sparking civil unrest or delegitimizing democratic outcomes. 'Uchaguzi Safe' addresses this critical vulnerability as a secure, citizen-centric platform engineered to crowd-source local polling station tallies and map security or administrative incidents in real time. By deploying localized anomaly detection models, the system filters out malicious duplicates, outlier reports, and coordinated misinformation campaigns. It presents a validated, rolling public average side-by-side with official Independent Electoral and Boundaries Commission (IEBC) results, establishing a dual-data validation ecosystem that enhances transparency, protects civic space, and fosters public confidence. Objectives: 1. Secure, Offline-First Citizen Reporting: Enable voters to securely document and upload localized polling results and incident reports (with image verification) via a highly optimized, low-bandwidth, offline-cached mobile application. 2. Machine Learning Anomaly Detection: Build an automated backend pipeline to instantly parse incoming reports, flag statistical outliers (using Isolation Forests), filter duplicate submissions, and detect coordinated misinformation. 3. Dynamic Geospatial Visualizations: Develop highly performant, live-updating interactive maps and voter turnout heatmaps using Mapbox to visualize national, constituency, and polling station levels in real time. 4. Dual-Data Verification Portal: Deliver a side-by-side public visualization contrasting consolidated crowd-sourced averages with declared IEBC results to promote transparent civic auditing. Scope of Work: - Cross-Platform React Native App: Low-bandwidth optimization, offline-first local SQL database caching, encrypted secure uploads, GPS geolocation pinpointing, and Mapbox geospatial overlays. - AI-Powered Python Backend: High-scale FastAPI web server processing inbound reports, utilizing Scikit-Learn (Isolation Forest and Z-Score algorithms) to filter malicious telemetry, and generating real-time rolling statistical averages per constituency. - Secure Admin & Moderator Web Portal: Role-based access control (RBAC) frontend built with React.js and Tailwind CSS for validated civil society observers to audit flagged outliers, verify incident photo uploads, and cross-reference manual tally sheets. - GIS Integration: Advanced mapping integration handling GeoJSON polygons of Kenyan constituency and ward boundaries with active heatmaps displaying voter turnout and security classifications. Key Deliverables: 1. Low-bandwidth optimized Figma UX/UI prototypes mapping comprehensive public and admin moderator journeys. 2. Production-ready, security-hardened iOS and Android mobile applications ready for play-store and app-store publishing. 3. Cloud-native Python/FastAPI backend coupled with PostgreSQL, fully integrated with Python-driven statistical outlier detection models, hosted on AWS. 4. Secure Web-based Admin Moderator Dashboard with multi-tier permissions for verified civic organizations. 5. Production-grade architecture documentation, GIS map-tile caching configurations, and a comprehensive security audit detailing end-to-end data encryption in compliance with the Kenya Data Protection Act.