Outfit Picker
an AI try-on app — upload your wardrobe and a photo of yourself, and generate a styled outfit preview.
Next.js · React · Tailwind CSS · FastAPI · Python · google-genai
why i built it
Outfit Picker is a full-stack app for visualizing outfits before you commit to them. you upload tops and bottoms into two browsable carousels, add a reference photo of yourself, describe a theme, and the backend generates an AI try-on image with Gemini. it's built as a complete product loop — a Next.js frontend talking to a FastAPI backend, file storage, and an external generation API with caching and a fallback path when the quota runs out.
selected details
- built a FastAPI backend with REST endpoints for wardrobe uploads, reference photos, and image generation
- integrated Google's Gemini API (google-genai) to generate a themed try-on image from a top, bottom, and reference photo
- cached generations by (reference + top + bottom + theme) so identical requests don't re-generate
- added a fallback mode that serves a placeholder output if the AI provider's quota is exhausted, so the pipeline stays testable end to end