Inspiration In Bangladesh, when local infrastructure breaks down (like a burst water pipe in Dhaka or a blown transformer in a remote thana), citizen complaints get lost in noisy Facebook groups or heavy bureaucracy. Meanwhile, government ministries are flooded with unstructured text and have no easy way to sort or prioritize these issues. I built NationBuild Civic Ledger to bridge this gap, transforming chaotic complaints into real-time, actionable public metrics.

What it doesNationBuild is a dark-mode, neon-accented civic tracking application:The Gateway: Users choose to file reports anonymously as a guest, or log in with an Email/Phone and Password to track their profile.Clean Statistics Dashboard: Shows strictly analytical metrics—Total Open, In Progress, and Fixed counters—alongside a ministry breakdown chart. It contains no messy text walls.Geographical Tree Navigation: In the "Issue Ledger," users filter complaints using a cascading dropdown of Bangladesh (Division , District , Thana). Typing is entirely eliminated.Semantic AI Triage: Google's Gemini 3.5 Flash instantly reads raw complaints, routes them to the correct ministry, rates severity and generates a clean headline.Social Democratic Prioritization: Citizens use Facebook-style "Agree" and "Disagree" buttons to vote, helping the state instantly identify high-priority community needs.

How we built itFrontend: React with styled interactive panels, glassmorphic cards, and a cyberpunk-inspired neon design.Backend & DB: A Node.js and Express API server synchronized with a real-time Firestore database schema for instant feed updates.AI Integration: Powered by the @google/genai SDK using gemini-3.5-flash configured to reliably parse complaints into structured JSON.We calculate a Weighted Public Urgency Index ($UI$) to help ministries prioritize

Challenges we ran into: Building the dynamic Bangladesh Geographic Tree Filter was our biggest challenge. Free-form text inputs caused query issues due to spelling mistakes and incomplete addresses. We solved this by mapping out a strict cascading dropdown structure of Bangladesh's administrative units, ensuring users can pinpoint their location with three simple clicks.

Accomplishments that we're proud of Bulletproof JSON Formatting: Forcing Gemini 3.5 Flash to reliably return clean, database-ready JSON keys without syntax errors. Zero-Noise Interface: Restricting the main dashboard to clean regional metrics, keeping it professional and uncluttered. Real-Time Syncing: Ensuring votes, database additions, and AI responses update instantly without browser refreshes.

What we learned Defensive UI Design: Replacing open text fields with cascading selection dropdowns dramatically saves database integrity and eliminates user typing errors. AI as a Routing Layer: Prompts require the same strict data validation, defensive rules, and safety bounds as traditional backend code.

What's next for NationBuild Civic Ledger SMS Gateway Integration: Allowing citizens in rural areas without internet to text complaints to be parsed by Gemini. Automatic Translation: Translating localized Bangla dialects directly into formal English summaries for state-level reports.

Share this project:

Updates

Submission history