Bengaluru Engineer Builds AI to Find and Fix Potholes

Every driver in Bengaluru knows the daily dance of dodging potholes, but one fed-up engineer decided to fight back with technology. Gaurav Sen built an AI-powered app that spots potholes, maps them, and emails the right government official to get them fixed.

The system works simply. Drivers mount their phone on the dashboard, activate drive mode, and the app takes a photo every second or two. When the AI detects a pothole, users can review the image and send an automated report directly to the responsible commissioner with one tap.

What makes this remarkable is how fast it came together. Using OpenAI's development tools, Sen built the entire system in a single day instead of the two months it would typically require. OpenAI commissioned the project to demonstrate how their technology can solve real-world problems without massive tech teams or long development cycles.

The app handles tricky conditions too. At night or in poor visibility, it inverts images and increases illumination to maintain accuracy. During rain, it can still detect potholes even when they're filled with water, though the confidence score drops slightly.

Why This Inspires

The pothole problem in Bengaluru isn't just annoying. Someone recently lost their life in a pothole accident, and some industries have even relocated from the city because of poor road conditions. Sen saw a crisis that touched millions of daily commuters and used his skills to create a practical solution.

The app is already open source on GitHub, meaning anyone can download and test it right now. Sen is actively encouraging the community to try it out, knowing that more real-world data will make the AI smarter and more accurate. Currently, he's tested it on about 15 to 20 routes around the city.

OpenAI's involvement shows they're thinking bigger than just Bengaluru. They see potential to scale this solution production-ready and deploy it across multiple cities. The goal is to showcase how accessible AI tools have become for solving civic problems.

The breakthrough isn't just about spotting potholes, which any person can do. The real innovation lies in documentation, accountability, and putting information into the hands of citizens who can demand action. By automatically identifying which contractor or department owns responsibility for each road section, the system removes excuses and creates transparency.

For now, the main limitation is that users need an OpenAI API key, which costs money. Sen is currently covering those costs himself but hopes community support will help scale the solution as more people join the effort.

What started as one engineer's frustration with bad roads could become a model for citizen-powered infrastructure monitoring anywhere in the world.