RunwAI
About this project
RunwAI is an AI-powered outfit generator developed as part of INIT Build, a collaborative student-led initiative. Users can upload photos of their clothing, which are automatically categorized using OpenAI’s CLIP model — a neural network that links images to descriptive prompts. The items are saved to a digital wardrobe, and outfit suggestions are generated using a custom rule-based algorithm that filters clothing by weather, occasion, and color. My primary role involved backend development and integrating AI functionality.
Key Features
- Wardrobe Categorization: Automatic wardrobe categorization using OpenAI’s CLIP model.
- Outfit Generation: Rule-based outfit generation based on user input (weather, occasion, color).
- Digital Wardrobe: Digital wardrobe management using Supabase.
- API Integration: RESTful API integration between frontend and backend.
Challenges
One of the biggest challenges was developing a functional AI component under a tight deadline. Setting up and fine-tuning the CLIP model locally, while ensuring it produced reliable results, required rapid learning and troubleshooting. We also faced technical issues with securely storing and retrieving user-uploaded images and metadata.
Learnings
This was my first experience working on a large-scale team project, which gave me valuable insights into collaborative development and agile workflows. I learned how to set up and run a local neural network, work with Supabase for cloud storage, and design an API that communicates effectively with a frontend. I also gained a better understanding of how AI can be integrated into user-facing applications.