My Projects
Education,personal & fellowship projects
Toxiscan
The app was developed using React, Next.js, JavaScript, MatureUI, Clerk and various libraries for barcode detection. We used the Open Food Facts API to retrieve product data and the barcode-detector library for barcode scanning using your native camera. The app is deployed on Vercel and hosted on our custom domain.
Toxiscan is a web application that uses the Open Food Facts API to provide information about food products regarding general information and their potential health risks and environmental impact. The app allows users to scan the barcode of a food item or search for it by name to view a detailed report. The report includes a product's ingredients, a nutrition score, a nova score, ingredient analysis, and a summary of the product's environmental impact. Users can also compare products and view recommendations for healthier alternatives.
Code WebsiteRateMyProfessorAI
The app was developed using React, Next.js, JavaScript, MatureUI, Pinecone, OpenAI, Firebase and Python. We used the Rate My Professor API to retrieve professor data and the OpenAI API serves as a chatbot that retrieves the needed information from Pinecone. The app is hosted on Vercel.
RateMyProfessorAI is a web application that uses the Rate My Professor API to provide information about professors and their ratings. The app allows Users to ask the chatbot questions about professors and courses, and the chatbot will provide answers based on the data from the RAG database created in Pinecone. retrieved from Rate My Professor. The chatbot uses the OpenAI API to generate responses and the Pinecone API to retrieve the needed information. We also created a scraping script that can take a rateMyProfessor link to scrape the site for the professor's or school's information which is then saved into the RAG database for general use for everyone.
Code WebsiteStudifyAI
The app was developed using React, Next.js, JavaScript, MatureUI, Clerk, OpenAI, Firebase, CSS and StripeAPI. We used the OpenAI API to serve as a chatbot that retrieves the needed information from the prompt the api was given and integrated a paywall and custom pricing plans using the Stripe API. The app is hosted on Vercel.
StudifyAI is a web application that uses the OpenAI API to generate flashcards based on the input given by the user. The app allows users to generate flashcards using OpenAI and save them to their account using Firebase. Users can also access premium permissions by subscribing to a monthly plan. The app uses the Stripe API to handle payments and subscriptions.
Code WebsiteToxiGuide
The app was developed using React, Next.js, JavaScript, MatureUI, Clerk, OpenAI, Firebase and CSS. We used the OpenAI API to serve as a chatbot that retrieves the needed information from the prompt the api was given to clearly provide information to the user. The app is hosted on Vercel.
ToxiGuide is a web application that uses the OpenAI API to provide information about Toxiscan and the purpose of the app. The app uses the OpenAI API along with the prompt with information regarding Toxiscan to generate responses.
Code WebsitePantryTrackz
The app was developed using React, Next.js, JavaScript, MatureUI, Firebase and CSS. The app is hosted on Vercel.
PantryTrackz is a web application that allows users to track their pantry items. Users can add items to their pantry, favorite items and even edit entries. The app uses Firebase to store user data and for user authentication.
Code WebsitePersonal Website
This website was created to showcase my projects and skills. It was built using HTML, CSS, and JavaScript, with a focus on responsive design. I used the Spectral template from HTML5 UP as a starting point and customized it to fit my needs. This project helped me improve my web development skills and learn how to use Git and GitHub for version control.
Machine Translation Comparative Analysis
This program was created for Professor Jonathan Gordon's Computational Linguistics class (CS366) as part of a final research paper. It provides a comparative analysis between direct translations (es-zh) and indirect translations (es-en-zh) using data from Tatoeba, an online community-built database of translations. We used transformers and HuggingFace libraries to select three models: one for direct translation and two for indirect translations. We fine-tuned these models with Tatoeba data and evaluated them using BLEU and ROUGE scoring metrics.
This project tested my team's teamwork and communication skills, as well as my time-management and leadership abilities. While the final product yielded favorable results, with more time, we could further improve the model's accuracy.
CodeVassar College Carpool App
This app was developed as a year-long project for Professor Rui Mereles's software design course. It allows Vassar students to create listings for trips, specifying their role (driver or passenger), trip time and date, pick-up and drop-off locations, and contact information. These listings are displayed on a dashboard accessible to all users, who can filter them by date, start and end locations, available seats, and role. Users can click on a listing to view detailed information. (This app is not published yet.)
Code