AI Skincare Scanner App (Computer Vision + TensorFlow)

OnSkin wanted to build an AI-powered skincare app that could scan a user's face, analyze skin conditions in real-time, and deliver personalized skincare routines. The vision was to replace guesswork with science, giving users dermatologist-level insights from their phone camera.

They needed a polished, high-performance mobile app that could handle complex image processing on-device while feeling effortless to use.

As CTO of M TECHUB LLC, I led the development of the full AI skincare scanner app for iOS and Android, combining computer vision with a clean, premium user experience.

AI-powered face scanning that detects acne, dark spots, wrinkles, texture issues, and skin tone unevenness

Real-time skin health score with detailed breakdown by zone

Personalized skincare routine recommendations based on scan results

Progress tracking with before/after photo comparisons over time

Product recommendations matched to individual skin profiles

Daily skincare reminders and habit tracking

Skin diary with notes, photos, and condition logging

Subscription model with premium features (unlimited scans, advanced analytics, dermatologist Q&A)

In-app community for skincare tips and user stories

Push notifications for routine reminders and progress milestones

Frontend: React Native (cross-platform iOS & Android)

Backend: Node.js with RESTful APIs

Infrastructure: AWS (EC2, S3, Lambda, Rekognition)

Database & Auth: Firebase (Firestore, Authentication, Cloud Messaging)

AI/ML: Custom computer vision model for skin analysis, TensorFlow Lite for on-device inference

Payments: Stripe and in-app purchase integration for subscriptions

Image Processing: OpenCV for real-time face detection and zone mapping

As CTO of M TECHUB LLC, I led the engineering team on this project. I owned the system architecture, mobile UI/UX direction, backend API development, AI model integration, and payment infrastructure. I worked directly with the ML team to optimize the skin analysis model for mobile performance and accuracy.

Website: onskin.com

Built a custom computer vision pipeline using TensorFlow Lite that runs skin analysis entirely on-device, keeping scan latency under 2 seconds

Implemented OpenCV-based face detection and zone mapping that segments the face into analyzable regions (forehead, cheeks, chin, nose, under-eyes)

Designed a progress tracking system that compares scan results over time and visualizes skin improvement trends

Built a personalized recommendation engine that maps detected skin conditions to product ingredients and routine steps

Integrated AWS Rekognition as a fallback for complex skin conditions requiring server-side analysis

Successfully launched on both App Store and Google Play