The rapid rate of urbanization in India has caused a gap between the citizens and the municipality authorities for public infrastructure management. Currently available grievance portals do not have capabilities such as complaint classification, prioritization based on rules, and accountability of contractors. In this study, we describe the development of a he proposed CIPS system using a web application that leverages machine learning on-device for complaint classification, scoring priorities based on set of rules, and cloud-based synchronization. The system uses Flutter (Dart) for cross-platform development, Google Firebase (Authentication, Cloud Firestore, Storage) as Backend-as-a-Service (BaaS), and TFLite-based natural language processing model as a part of the Edge AI process for categorizing the complaint into pre-defined categories of infrastructure. Our solution includes a multi-user role-based architecture where roles such as Citizens, Authorities, Contractors, and Administrators are available to cover all stages of complaints from reporting till contractor assignment, work completion, and spatial analytics. Experimental analysis shows an ML inference latency of around 120ms, notification delivery latency less than 2 seconds on Firestore, and a Lighthouse Performance score above 85.