TeamUP

Inspiration

Coming out from across the country, we were struggling to find a team for the UC Berkley AI Hackathon. Although the in-person team formation process is helpful, it takes away from our precious hacking time and delays the ideation phase -- we don't want to miss a single minute that could be spent coding! This is how and why we were inspired to create TeamUP, an application to help streamline the team formation process online by matching hackers with the power of AI tools via an interface similar to Tinder.

What it does

TeamUP is designed to help match hackers with potential teammates with based on a combination of technical and soft skills. When creating your hacker profile, you'll be asked to list some of the libraries, frameworks, and tools you're experiences in so that our recommendation-system can try to suggest other hackers with skills that complement yours based on the theme of the specific hackathon. However, as often is the case with teamwork, we also want to make sure that you'll get along with your potential teammates which is why TeamUP also analyzes your "about" section to match you with people who will likely create the optimal collaborative space. This makes sure teams are well-rounded and equipped with a variety of different skillsets to create the best projects possible.

How we built it

For the web app, we used the React (node.js) framework for the frontend and Express framework for the backend. We hosted the database in MongoDB Atlas and used AWS to host our backend APIs. For the matching system, we utilized HuggingFace's Inference module, specifically the deepset/roberta-base-squad2 model.

Challenges we ran into

Finding a team! We spent quite a lot of time finalizing our team due some hiccups in communication... but we eventually got there. On the technical side, one of the biggest challenges was integrating the frontend and backend services because there were so many components. It also took us some time to get the AI model working with our react app -- it was a time crunch for sure!

Accomplishments that we're proud of

We are proud of our web application and integrating AI for the matching system. It was a time crunch, but each and everyone of us learned something new which is the greatest reward.

What we learned

We learned that full-stack development is fun but challenging at times. We also gained an appreciation for the ideation and prototyping phases because even if you put in more time there, it will make the rest of the hacking process smoother since everyone has a clear idea of what they need to be working on. Additionally, we got hands-on experience and enhanced our skillset in a variety of technologies, including MongoDB, Express, React, AWS, and more. All-in-all, this was another great opportunity to see how we can harness the power of AI to enhance our daily lives through accessible applications.

What's next for TeamUP

Although TeamUP is initially designed for AI-themed hackathons like this one, we would like to broaden the scope to all annual collegiate hackathons (themed or not!) so that we can find better skill matches based on the specific hackathon(s) the user selects. We would also like to improve our hacker matching system to make it more sophisticated, such as by noting post-hackathon feedback so that it can get a better sense of your teamwork preferences. This would require more research into just how recommendation-systems work from the ground up, in addition to making our app more powerful to personalize the experience for every hacker.