Inspiration
Micheal Heilman. For my senior thesis I'm trying to make a robot automatically generate multiple choice questions, and this project was an experiment to see if the current system I built could be reformed to instead generate flash cards for studying purposes. Micheal Heilman, was the last researcher before me to really take on this task, and my system barrows heavily from his previous work.
What it does
Flashtastic is a website/AI where you can submit a url/pdf/raw text and within a short minute or two, it will provide you with flash cards based off of the material. It does this with a fair amount of NLP wizardry that determines which word phrases are likely to be testable, and just writes flash cards for those words. Built into the site is an easy way to filter any noisy suggestions, so you can get to studying ASAP.
How we built it
The system was built with a multitude of languages and frameworks, largely due to the fact that many micro components were necessary to really get this to work, and they all came from different developers with different language preferences. So it's pretty much a web of flask/node/spark servers. It's hosted on google cloud and the website if built with react.
Accomplishments that I'm proud of
I'm extremely surprised by how easy it was to reconstruct the existing system to serve this experimental purpose. Also the results were better than expected.
What I learned
Sometimes the naive solutions is the best way to go ( for a hackathon that is. )
What's next for Flashtastic
This Hackathon strengthened my passion to take this application beyond just the laboratory. Currently the system is closed sourced, but If I end up winning the 5k server prize, it wont be, and optimistically a running public service would be available within 1-2 months.

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