The devastation caused by Hurricane Harvey and Irma inspired us to help affected denizens become aware of the resources available to them.
What it does
Bridges the gap between charity organizations and people in distress by making it easy to locate and offer aid.
How we built it
We used Angular and Flask for the front-end web application, and MongoDB for the backend database. Our search algorithm was based on word2vec, using pretrained data from Facebook's fasttext, and linked with gRPC.
Challenges we ran into
Formatting the word vector embeddings proved very memory intensive and our machines had to use swap memory, slowing us down considerably in the backend development. We experienced difficulty integrating the front-end with the back-end using gRPC, and were ultimately, unable to finish this link.
Accomplishments that I'm proud of
Learning about the efficiency and power of word2vec to leverage the spacial embedding of words to make a smarter and faster database search was super nifty.
What we learned
The power of NLP techniques such as word2vec's skip-gram model. We were also amazed by the power of Angular and Flask to structure and speed up development of the front-end.
What's next for Relief
Integrating Google Maps to provide directions (possibly with alternative routes if certain roads are inaccessible).
I am grateful for the contributions of Bijan Varjavand, Gabriel Villasana, and Richard Guo, without whom, this project would not have been possible. I am also thankful to Brandon Duderstadt for his suggestions on how to improve all aspects of the stack.