Live project: BrihanMumbai Fix
I got tired of hearing the same sentence over and over again: "Nothing changes."
Potholes stay open. Garbage piles up. Streetlights remain dead. Citizens complain in WhatsApp groups, Instagram comments, and chai-time conversations, but there is no clean, structured path from problem spotted to problem resolved.
So I built BrihanMumbai Fix: an AI-powered civic complaint platform focused on Mumbai, designed to make issue reporting faster for citizens and issue handling clearer for authorities.
This post is the full story: the social problem, why it matters, why I decided to build this, how I built it, and exactly how the workflow works end-to-end.
The Problem I Wanted to Solve
Mumbai is full of high-frequency civic problems:
- Potholes that damage vehicles and cause accidents.
- Garbage and sanitation issues that create health risks.
- Water leakage and drainage issues that get worse during rain.
- Broken streetlights and electrical hazards that affect safety.
The core problem is not just the issue itself. It is the reporting friction:
- People don’t know where to report.
- Reports are often unstructured and missing details.
- Authorities receive noisy or incomplete complaints.
- Citizens lose trust because they cannot track progress clearly.
That gap is exactly where BrihanMumbai Fix sits.
Why I Decided to Build This
I didn’t want to build “another app.” I wanted to build a practical civic utility.
The idea was simple:
- Let citizens upload a photo in seconds.
- Use AI to classify the issue and generate useful context automatically.
- Route the complaint with ward and department-level structure.
- Give admins a focused dashboard to triage and update statuses.
- Keep citizens informed through dashboard + community feed visibility.
In short: reduce ambiguity, reduce effort, and increase accountability.
What I Built (Tech Stack)
BrihanMumbai Fix is a full-stack project split into frontend and backend.
Frontend
- Next.js 16 (App Router)
- React + TypeScript
- Tailwind CSS + shadcn/Radix UI
Key pages include:
- Public landing page
- Login / register
- Report issue flow
- User dashboard (personal complaints)
- Community feed
- Admin login and admin dashboard
Backend
- Flask monolith API
- MongoDB for users + complaints
- JWT auth for user/admin roles
- Cloudinary for image storage
- Gemini Vision for image analysis
- Groq for fallback analysis and formal complaint text generation
How the Product Works (End-to-End Workflow)
Here is the real workflow from the moment a citizen spots a problem.
1. Citizen Authenticates
A user registers or logs in, then gets access to protected flows like /report, /dashboard, and /feed.
2. AI-Assisted Complaint Creation
Inside the report flow:
- User uploads an image of the civic issue.
- Frontend sends it to the backend analysis endpoint.
- Backend uploads image to Cloudinary.
- AI model analyzes the image and returns:
- Issue type
- Severity
- Description
- Department mapping
- Confidence
This reduces manual typing and helps normalize complaint quality.
3. Location + Ward Context
User confirms location, ward, and optional coordinates/details. This is critical because municipal action depends on jurisdiction.
4. Complaint Submission
Backend validates fields, stores complaint in MongoDB, and can generate a formal complaint-style text using LLM assistance.
5. Citizen Tracking
Users can view their complaints in the dashboard, including status progression and timestamps.
6. Public Community Visibility
The protected community feed shows recent complaints with ward, issue type, severity, and status. This creates social transparency and helps citizens see that they are not alone.
7. Authority/Admin Operations
Admins log into a dedicated portal and can:
- View and filter complaints
- Open complaint details
- Update status as work progresses
- Monitor complaint volume via stats
This makes the operational side as important as the reporting side.
Architecture Decisions That Mattered
A few decisions made this project actually usable rather than just demo-friendly:
- Role-based auth model for user vs admin paths.
- Middleware route protection for report/dashboard/feed/admin access.
- AI-first report assist so complaints are structured from the start.
- Ward-aware complaint metadata to improve handoff to local authorities.
- Separation of concerns between citizen UX and authority triage UX.
Challenges I Faced
This wasn’t just “connect frontend to backend and done.”
Real issues included:
- Syncing frontend auth state with middleware-verified server state.
- Managing clean logout/login transitions between admin and user roles.
- Preventing repeated AI analyze submissions in the report flow.
- Keeping API behavior consistent across protected routes.
Building civic systems forces you to care about reliability because user trust is fragile. If the flow breaks once, people stop reporting.
What This Project Actually Solves
BrihanMumbai Fix does not magically fix every pothole. But it fixes a major bottleneck:
- It turns scattered complaints into structured civic data.
- It cuts reporting friction for citizens.
- It gives admins a single operational surface.
- It creates status visibility and stronger follow-through.
That is how digital systems can make cities incrementally better: not by hype, but by reducing everyday friction.
Final Thoughts
This project started from frustration, but it became a serious full-stack civic tech build.
I learned that AI is most useful when it supports humans in messy real-world workflows, not when it replaces them. In this case, AI helps classify, structure, and accelerate civic reporting, while citizens and authorities still drive the actual outcomes.
If you’ve ever been angry about a city problem and felt powerless, build a tool. Even a small one. Systems change when someone decides to replace complaining with architecture.
What’s Next
Planned improvements for BrihanMumbai Fix:
- Better anti-spam and rate-limiting controls.
- Richer complaint timeline tracking.
- Ward-level analytics and heatmaps.
- Stronger moderation and verification workflows.
- A tighter mobile-first reporting experience.
If this resonates with you, start with one neighborhood-level problem and build from there. Civic tech gets powerful when it is local, specific, and relentlessly practical.
Disclaimer
This project is created for educational purposes and as a concept demonstration only.
It is not an official government platform and is not affiliated with, endorsed by, or connected to any government authority.