💡 Inspiration

Traditional skin analysis tools rely solely on superficial visual AI data, completely ignoring the internal physical comfort of the user. We created Skin Comfort Loop to bridge this critical gap, ensuring that what the AI sees perfectly aligns with what the user actually feels.

After witnessing how standard routines can trigger silent irritation, we wanted an app that dynamically crosses biometric data with raw user feedback.

🔎 What it does

Skin Comfort Loop synchronizes real-time visual biomarkers from YouCam AI with daily user comfort micro-logs (tracking tightness, sensitivity, and texture).

When the AI detects visual improvement but the user reports decreasing physical comfort, our system flags this data divergence and automatically triggers tailored clarification prompts to adjust the skincare routine before irritation breaks out.

⚙️ Core End-to-End Workflow:

• Scan ➔ User uploads a photo for visual biomarker processing.
• Log ➔ User submits their physical comfort index via a 3-tap form.
• Analyze ➔ System plots a real-time dual-curve graph mapping AI Score vs. User Comfort.
• Detect ➔ Custom logic flags any data divergence trends.
• Intervene ➔ Active clarification fields display instantly to adjust routine parameters.

🛠️ How we built it

The application leverages a robust server-side architecture to process telemetry data cleanly and securely:

• Frontend: Built with React, TypeScript, and Vite for a highly responsive, mobile-first interface.
• Backend Functions: Tailored server functions manage private pipeline processes and shield upstream networks.
• Third-Party AI Integration: Fully integrated the Perfect Corp YouCam Skin Analysis API (v2.0) through secure backend-to-backend calls.
• Database & Cloud Storage: Hosted entirely on a secure cloud-managed database leveraging isolated buckets and strict user-scoped Row-Level Security (RLS).

⚠️ Challenges we ran into

We faced significant integration challenges aligning with the YouCam external API contracts, resulting in sequential 403 Authorization and 404 Endpoint errors due to strict signed URL headers and path formatting.

Debugging these silent fallbacks required migrating the pipeline from v1.1 to v2.0, removing conflicting manual header mutations, and standardizing proper payload handling for live production validation.

📊 Accomplishments that we're proud of

• Zero-Leak Token Architecture: Implemented a secure backend isolation layer that protects sensitive provider keys from client-side exposure.
• Production-Grade API Fixes: Decoded and resolved deep REST protocol conflicts directly against complex signed URL specifications.
• Robust Fallback Engine: Deployed an automated, predictable fallback handler ensuring zero downtime for the user interface during upstream API timeouts.

📖 What we learned

We learned the extreme importance of strict API contract verification, proper handling of custom formatting within signed cloud storage architectures, and the necessity of isolating third-party microservices behind protected server layers to safeguard credentials.

🚀 What's next for My Skin Loop

We aim to scale the platform into a comprehensive B2B dashboard integration, allowing skin coaches, estheticians, and dermatologists to seamlessly monitor client divergence logs remotely without having to manually sift through raw visual data.

Built With

• backend
• cloud-storage
• functions
• git
• github
• perfect-corp-api
• postgresql
• react
• rest-api
• row-level-security
• supabase
• typescript
• vite
• youcam-ai