BlockWatch

Inspiration

This is our video Presentation link: link

We were inspired to create BlockWatch as a tool to help citizens navigate through NYC safely and to be aware of dangerous situations.

What it does

BlockWatch is a website that helps people navigate New York City more safely by combining real-time AI with hyper-local data. Users describe their travel plans, and our system guides them to provide important context like time, location, and who they are. We use OpenAI to analyze the user’s needs and select which safety data is most relevant. MongoDB’s vector search finds the top matches from NYC crime and 311 data, and OpenAI generates a simple safety summary for the user’s route or destination.

Users then see interactive, AI-generated safety insights for their trip on our custom React site, featuring a mapbox-gl map of NYC. The sidebar updates instantly with relevant info, and users can also search any ZIP code to explore local crime trends and patterns by precinct and time of day.

How we built it

Technologies used: React (frontend/UI), mapbox-gl (interactive map), MongoDB (crime data storage and vector search), OpenAI (AI reasoning and summaries).

Accomplishments that we're proud of

We're proud to have built this app and learned more about the capabilities of Mongodb.

What's next for BlockWatch

BlockWatch needs to increase it's data scope to include a larger area of NYC and potentially America, while increasing its functionality to recommend new travel routes and other plans for users to safely get to their desired location.

Built With

Submitted to

AI in Action

Created by

I worked on the backend connecting MongoDB to the react website and sending the Openai summaries to the frontend and overall website flow

Ray Ray
I built the front-end using React, HTML, and CSS. I integrated Mapbox GL to render the map and wrote the code for interaction. I also contributed to the overall design and layout.

Shasuna Lin
I pulled, cleaned, and merged NYC crime and 311 data for 2024, linking each incident to its precinct, ZIP code, and police tour while dropping unusable records to keep things fast. After digging deep into public datasets and fixing mismatches, I created a ready-to-query dataset that we then uploaded to MongoDB and processed for vector search, letting BlockWatch serve hyper-local safety insights by area and time of day.

Daniel Foulen
Erica Bouloute

Updates

Ray Ray started this project — Jun 17, 2025 04:21 PM EDT

Leave feedback in the comments!

Log in or sign up for Devpost to join the conversation.