RoadAlert AI

Inspiration Every Indian has seen a pothole and complained about it to nobody. The real problem isn't that authorities don't care — it's that most citizens never file an official complaint because the process is complicated, confusing, and time consuming. We wanted to fix that. What it does RoadAlert AI lets any citizen upload a photo of a civic issue, enter the location, and instantly receive a fully formatted official complaint letter ready to send to the Municipal Commissioner. It identifies the problem, assesses the danger level, names the responsible government department, and generates the letter — in English or Hindi. How we built it We built RoadAlert AI using Google Gemini's vision API for image analysis, Python and Streamlit for the frontend, and python-docx for generating properly formatted Word document complaint letters. The entire app runs from a single file with no database or complex backend. Challenges we ran into Getting Gemini to return consistently structured output was the biggest challenge. We solved this by engineering a strict system instruction that forces the model to respond in a fixed format every time, which we then parse to populate the complaint letter cleanly. Accomplishments that we're proud of A fully working end to end pipeline — from raw photo to downloadable formal complaint letter — built and deployed in under 3 hours, solo, on an unfamiliar PC. What we learned Prompt engineering is as important as the code itself. The quality of Gemini's output is entirely determined by how precisely you instruct it. We also learned that the best hackathon projects solve a problem every judge has personally felt. What's next for RoadAlert AI Direct integration with municipal corporation complaint portals so the letter gets submitted automatically. Multi-language support beyond Hindi. A mobile app so citizens can report issues on the spot. And a public dashboard showing reported issues on a city map.

Built With

• gemini
• google
• python
• streamlit