Waypoint | Devpost

Inspiration

While driving down here and looking at all the challenges, we noticed the one titled "Climate Tech Sustainability Challenge" and were heavily inspired by it. Thinking back to Hurricane Milton last year, we realized the severity of the danger disasters bring onto Tampa and wanted to create something to alleviate the aftermath of such devastation.

What it does

Our project provides a central system that connects survivors to responders. Survivors are able to use the chat feature to report distress, which creates a heatmap on the map responders can view. Survivors are also able to see the current responders in their area on their map. Responders are able to query quick data about the current situation, along with being able to generate reports on specific heatmap regions and broadcast important messages to the users.

How we built it

For the backend, it was built using a single FastAPI server running both a WebSocket and REST API to handle different functionalities such as messages, queries, and live location sharing. Each user and message was stored in a Postgres database with extensions for geolocation and vector databasing, allowing us to use it as a RAG system for Gemini to query and answer questions about specific regions and situations. This backend was coupled with a frontend written in React and powered by Vite to display the different scenarios, portals, messages, and interactions between responders and survivors.

Challenges we ran into

We ran into various challenges both technically and in implementing this idea into reality. In the backend, we had to go through various optimizations on how we sent and stored the data as vector embeddings so we were accurate in retrieval. We spent the majority of our time on this, as it was significant for our app to function properly and to serve information quickly. On the frontend, we ran into issues when visualizing the map and various layers on top of it. We had to work carefully to display the heatmaps and the pins layer by layer.

Accomplishments that we're proud of

For most of us, this was the first time building a RAG pipeline from the ground up. So we are proud of the fact that we were actually able to integrate it with the app itself and create successful interactions across multiple parts of the app.

What we learned

We learned a lot about RAG and its interaction with a full-fledged app. We understood that it doesn't solely exist as a data pipeline but as part of the app, and we demonstrated this through how the full app worked. We are also proud of pulling off this idea and being able to implement something that may have significant value in the real world.

What's next for Waypoint

For Waypoint, we want to be able to implement voice features where responders are able to make queries more quickly and get information faster. We also want to focus a bit more on accessibility by allowing the app to be used by anyone at any time.