Atelier

View live

Design studio for fashion designers that turns a photo of any garment into a ready-to-cut sewing pattern, a ghost-mannequin product shot and a virtual try-on.

Atelier landing page — “Studio-Ready Fashion From a Single Photo”
Atelier — the cost of ghost mannequin photography, solved by the Ghost Mannequin tool
Atelier — drafting patterns from scratch, solved by the Pattern Generator
Atelier — weeks-long iteration cycles, solved by all three tools together

Atelier is built for fashion designers, automating the slow, manual parts of their work. It drafts the sewing pattern for any dress the way a tailor would, with each piece drawn out and ready to cut, and it also produces ghost-mannequin product shots and AI virtual try-ons that show a client wearing the garment. Designers track jobs live, save results into projects and export them as PDF or ZIP. I built a secure Go (Echo) backend on PostgreSQL with an asynchronous job system built with Go channels and a pool of worker goroutines. Workers claim jobs atomically in the database, so no two workers process the same job, and pending jobs resume after a restart. The backend is deployed on Cloud Run with Cloud SQL, GCS and Secret Manager. The Next.js frontend, including a 3D mannequin viewer built with React Three Fiber, runs on AWS ECS Fargate behind CloudFront. Infrastructure is provisioned with Terraform, and deployments are automated through GitHub Actions.

Technologies

  • Go
  • Echo
  • PostgreSQL
  • JWT
  • Next.js 16
  • React Three Fiber
  • Cloud Run
  • Cloud SQL
  • Cloud Storage
  • Secret Manager
  • AWS ECS Fargate
  • CloudFront
  • S3
  • Terraform
  • GitHub Actions
  • React
  • TypeScript
  • Tailwind CSS
  • TanStack Query
  • Radix UI
  • Framer Motion
  • Three.js
  • Gemini
  • Replicate
  • AWS
  • GCP
  • Docker
Next