AI Skin Decode – AI-Driven Skin Analysis & Recommendation Platform for Acnestar (Mankind Pharma)
AI-driven skin analysis and product recommendation platform for Acnestar, a dermatology brand by Mankind Pharma
Project Overview
AI Skin Decode is an AI-powered skin analysis platform developed for Acnestar (Mankind Pharma) that leverages TensorFlow.js and computer vision models to analyze facial images, detect acne patterns, and generate dermatologist-backed product recommendations. I worked on building the complete AI-driven frontend experience, focusing on real-time camera capture, multi-angle face analysis, ML inference workflows, and clear presentation of medical-grade skin insights.
Project Gallery
Loading...
Featured Screenshot
Landing screen introducing AI-powered skin analysis with a guided CTA
Loading...
Loading...
Loading...
Loading...
Loading...
Problems Solved
Implementing reliable real-time face detection using TensorFlow.js in browser environments.
Ensuring accurate face alignment and angle capture for ML model inference.
Reducing false positives caused by lighting conditions and camera auto-enhancements.
Designing a guided UX to minimize user error during multi-step image capture.
Translating raw AI detection results into clinically understandable skin conditions.
Maintaining user trust and medical credibility in an AI-assisted health product.
My Responsibilities
Architected and implemented the AI-powered frontend using Next.js and React
Integrated TensorFlow.js and BlazeFace for in-browser face detection and analysis
Built guided selfie capture flows for front, left, and right facial profiles
Implemented real-time visual overlays for lighting, positioning, and angle validation
Designed analysis result views mapping detected acne types to product recommendations
Optimized performance of ML inference to run smoothly on mid-range mobile devices