Welcome to the SMIDGen Project, part of the Sensify Lab at the University of Delaware. This project focuses on developing user-friendly tools to assist scientists in collecting, filtering, and understanding their data, with the long-term goal of improving data quality and reproducibility. As an undergraduate researcher, my contributions to this project have spanned UI & Front-end development, backend features, and AI-powered labeling.
Technologies Used: HTML, CSS, Bootstrap
I was responsible for redesigning the user interface of the SMIDGen website to improve usability and visual appeal. The redesign aimed to create a more intuitive experience for users accessing the application.
Technologies Used: Node.js, Express.js, Firebase
I worked with another undergraduate researcher to develop a feature's front-end and connect it to the backend that allows admins to manage user roles within the team. This feature enables administrators to:
- Add new users
- Update or change user roles
- Remove users and clean up the team and project's database
Technologies Used: GPT-4 API, Prompt Engineering
I participated in a summer research project where I helped engineer prompts using GPT-4 on Google Colab to assist in labeling articles. This was part of a broader study focused on AI-human interaction dynamics. The key goals of this research were to explore:
- How to use AI to suggest more accurate article labels
- Improve human trust in AI-generated suggestions by determining the best interface design for users
Additionally, I contributed to the literature review, created figures, and assisted in writing a research paper that was submitted to the CHI 2025 conference.
Building on the interfaces developed during the summer research, we are now implementing this feature into the SMIDGen web app. The goal is to use AI to suggest labeling while providing AI explanations that justify why it made certain choices. This ongoing project heavily focuses on:
- UI development
- OpenAI API integration
We are excited to see how this feature will enhance the user experience and contribute to more accurate and understandable AI-driven insights.