Case study · Beauty-tech
Venus — an AI skincare coach on iOS and Android
Venus is a K-beauty program in an app: scan your face, get a plan from AI and Seoul skincare experts, follow a routine, and track how your skin changes. I joined at the MVP stage and owned its engineering for the next 18 months as it grew into a subscription product.
- ≈10,000
- users
- iOS + Android
- live on both stores
- 18 months
- Apr 2025 – Sep 2026
- v1.4.x
- shipped release after release
My role
Joined at the MVP stage and led engineering for the next 18 months — mobile app, backend, AI pipeline, payments and releases.


What I worked on
AI skin analysis
Guided selfie capture and upload, then an OpenAI vision pipeline with engineered prompts and a structured output schema. It scores each concern — acne, pores, eye bags, redness, oiliness — and writes the detail cards users read.
Personalized routines and product search
A LangChain pipeline from diagnosis to a personal routine, with Pinecone retrieval over products and skincare knowledge, ingredient checks, and a cache in front of the model to keep inference costs down.
Daily check-in and progress
The habit loop that brings people back: a daily check-in with a weekly score calendar, and before/after comparisons so users can see their skin change.
Subscriptions and expert consultations
Venus Pro subscriptions through RevenueCat and StoreKit, onboarding and paywall flows, and intro consultations with Korean skincare experts.
Growth and attribution
Amplitude analytics, Facebook and TikTok SDK attribution, Branch deep links and campaign tracking, so the team could see which channels brought paying users.
Backend, infrastructure and releases
FastAPI and Postgres on AWS, with image uploads to S3 served through CloudFront, Docker and CI. Expo builds shipped continuously through TestFlight and Google Play.
Stack
- App
- React NativeExpoTypeScript
- Backend
- FastAPIPostgres
- AI
- OpenAILangChainPinecone
- Payments & growth
- RevenueCatStoreKitAmplitudeBranch
- Infrastructure
- AWSS3CloudFrontDockerCI
Results
- Live on the App Store and Google Play
- About 10,000 users
- Subscription business in place: paywall, Venus Pro and expert consultations
- Shipped release after release through version 1.4.x

More detail and screenshots on my portfolio case study ↗
What this means for your app
Venus has the parts most AI apps need: a model call people trust, a reason to come back every day, and a subscription that pays for it all. The free 3 days are where those parts start.
Contact
Start with your idea
Send your idea, get a technical plan
Describe the app and leave your email. I'll reply with a technical plan — the stack, what the free 3 days will build, and what it would cost to finish.