I was a nominee a Repl.it's award at Stanford's 2023 Treehacks Hackathon!
In February of 2023, I was invited to compete at Treehacks, Stanford's annual Hackathon. Along with a team of 3 others, with two of us attending our first Hackathon, we created a smart AI shopping assistant utilizing OpenAI's JavaScript API.
My contribution to the project was a user-friendly UI which the user could converse with the bot. I also created a sales presentation which we used to pitch our product to judges, includes representatives from OpenAI, Google, and Repl.it. My work was entirely frontend, being a user-friendly chatbox to give OpenAI's API prompts and recieve responses. The purpose of this work was to make asking questions as simple and straight-forward as possible for the customer.
My work features a button that appears in the bottom right of the client's screen. When clicked on, a retractable chatbox expands to fill the right half of the screen. Here, the user may ask any question regarding a product on the website. Any question that is "undefined" or an empty string will not cause the AI to respond. Only when a valid input is provided, the AI will respond, creating a new reponse textbox with a respective profile picture. Messages may be sent with the send button, and the chatbox may be closed with the close button.
Prior to the event, I had no team or knowledge of what a Hackathon would be like. I was very excited, but admittedly nervous. Around two weeks before the start date, I began reaching out to fellow hackers. Together, we managed to form a team of 4 from across the country, with none of us having met before. It also happened to be one of my fellow teammates first Hackathon as well!
We went through many ideas and subjects of which to code, but finally settled on a Smart AI that could effectively eliminate the need for customer support. The program would utilize OpenAI's API, and would have its own website to demonstrate both the administrative and customer experience.
With Treehacks being 36 hours, we set to work right away. One of my teammates, Adi, focused on the backend features and figured out how to train OpenAI's API with custom data. He then focused on how to permit the AI to provide for that data. My other two teammates, Alyssa and Sabrina, focused on creating the frontend homepage of the website, with a navigation bar and displayable product lists.
As I mentioned earlier, I focused on creating the actual interface that the user could discuss with the bot inside of. Fortunately, after over 20 hours of straight coding---and much testing---our program came to life after linking our code.
We were even nominated for Repl'it's award of the best usage of their platform, but unfortuntately did not win. Better luck next time ;)
