Right to Know

Inspiration

The nation sees a large number of police misconducts, and some misuse their powers. Also, a major population of citizens are unaware of their rights and the laws that protect them.

"Most misconduct involves routine infractions, but the records reveal tens of thousands of cases of serious misconduct and abuse. They include 22,924 investigations of officers using excessive force, 3,145 allegations of rape, child molestation and other sexual misconduct and 2,307 cases of domestic violence by officers." - US Today.

Some cases of excessive force or misinformed citizens can be avoided by helping citizens know their rights when faced with a tense situation (interaction with law enforcement). We built a mobile web application to help citizens know their rights and mutually de-escalate situations.

What it does

The app listens to the conversation between the citizen and the law enforcement officer. Based on the ongoing conversation, the app constantly suggests the person with prompts, telling them the possible ways to handle the situation. The user's location is also used to review laws pertinent to the particular state and county.

How I built it

We created a simple Python Flask application. The connection between the backend and frontend was handled by using SocketIO. The audio was transcribed and sent to Gemini API with the required supporting prompts, and the suggestions were shown to the user.

Challenges I ran into

Using SocketIO stream was a pain while streaming the audio. The library is not widely used, and hence, we couldn't debug some issues. We moved to an implementation where we segmented the audio and sent chunks to the backend.

Accomplishments that I'm proud of

Got an opportunity to use AI to create an impactful application that helps citizens be informed in conversations with law enforcement officers.
Got to integrate LLM APIs for the first time