Hairstyle Vision
Overview
AI hairstyle consultation for barbershops. The barber photographs the customer in the chair, the system simulates the styles that branch actually cuts on that customer's own face, and both sides choose from the same screen before any hair is touched.
Responsibilities
- Built the whole product solo — Go backend, React front end, PostgreSQL schema and deployment.
- Ran hairstyle simulation on a self-hosted open-source image model rather than a third-party API, so customer photos stay inside the system.
- Designed the QR-paired session flow: the studio screen opens a session, the barber scans it and shoots with their own phone camera.
- Added a capture gate that checks face, hair and lighting before a photo is sent for simulation.
- Scoped the hairstyle catalogue per branch, so each location only offers the cuts its own barbers serve.
- Persisted every session with its before photo and simulation as a reference for the next visit.
Result
- Customer and barber agree on one cut from the same screen before any hair is touched
- Each style is previewed on the customer's own face rather than a reference photo of someone else
- Every branch offers only the cuts it actually serves, set from its own catalogue
- Sessions are kept with the before photo and simulation, giving a repeat visit a concrete reference
- Customer photos never leave the system — simulation runs on self-hosted inference
Screenshots
