Global Investment Research Division
Created a RESTFul API using Java SpringBoot that would parse through a research document, send the text to our first Python Flask server which would use the spaCy Natural Language Processing (NLP) library to get the named entities within the research document, then send those entities to our second Python Flask server that used an Entity Resolution (NER) Algorithm to match the spaCy entities to our internal database. Our team then integrated our service with the research editor to inject found named entities as suggested tags into the editor. The work done on the editor required building a UI using CoffeeScript and jQuery, with some of the original prototypes using Angular.JS.
Givback was a charity donation webapp that would allow a user to round up the cost of their purchase to the nearest dollar and donate that to charity. Givback had a built in charity information search that would display the aggregate financial data from a charity in a meaningful way. Read More
This webapp was created with Materialize and jQuery for our frontend and Node.js with Express.js for the backend. We used the Plaid API to securely handle bank information and transactions and OrgHunter API for the Charity finacial information.
The structuring of the charity financial information was what made Givback stand out from any project that may be like it. Not only is it about giving back to people and communities that may need it but it is also about making sure the user is informed about how their contributions would be spent. Seeing that their contribution makes the biggest impact it can.
GSTR (Gesture) is a web-based application that helps bridge the gap in non-verbal communication. Using Clarifai’s Custom Training technology, GSTR recognizes sign language from visual inputs and then gives feedback and translation information. Read More
This webapp was created using Materialize, jQuery, WebcamJS for the frontend and Node.js with Express.js for the backend. Clarifai API was used for the image recognition and the custom model was trained using a node script that would send over 3,000 images to clarifai with the correct concept
Using Clarifai to train a custom model instead of creating our own neural network allowed for us to make a quick proof of concept for what an app like GSTR can do. For GSTR to work at it's full potential we would need to create our own model and have a much bigger set of material for the network to learn from. That is something that is planned for the future
JASM (Just Another Smart Mirror) is a mirror with built in voice commands. The original project that was built for MangoHacks had more features than the version that is currently available. The current version has "Jas, Chill" which opens netflix and plays "Careless Whispers" by George Michaels, "Jas, Mixtape" which opens up my this website, "Jas, Help" which opens up the options menu, and "Jas, Home" which hides the options menu.
The mirror was actually a 16" x 17" 2 way piece of acrylic with a 19" monitor behind it running our webapp that was hosted off a raspberry pi apache server. The original webapp was very crude and not responsive to different screen sizes so I have started to remake it using bootstrap and the knowledge I've gained since first making that project
GSTR uses Clarifai through a custom visual recognition model that the GSTR team trained on sign language to recognize signs and gestures. It then responds in two different modes - educational and translation. Educational mode provides feedback confirming correct implementation of the sign. Translation mode gives the text and audio definition corresponding to the sign.
Nearly 250 students from all over the state and even across the country spent March 4-6 at Tech Station for the first large-scale, student-led hackathon hosted at FIU. They whipped up more than 30 projects, including a Smart Mirror that can bring up a live news feed and information about the weather and maps; an app that helps connect communities face-to-face; and a role-playing game that allows users to learn Spanish, Portuguese or French.
Among the winners of the Hackathon were an open sourced API for facial recognition software, a city wide umbrella rental service via kiosk, and a mirror that could display data from your iOS device that answered to Siri commands using a voice recognition API.