Clients arrive with vague references while barbers translate those references through face shape, hair behavior, and maintenance constraints. That ambiguity costs time and confidence before the cut even begins.
← Portfolio overview02 / 04
Appearance intelligence
BAIR
See the haircut that fits before you step into the chair.
A mobile-first haircut and skincare product that turns face landmarks, explainable scoring, and image editing into a clearer conversation between clients and barbers.
- Role
- Founder, researcher, product designer, and engineer
- Year
- 2026
50+Barber interviews
6Ranked recommendations
10 minReported appointment reduction
The brief
BAIR uses MediaPipe facial landmarks and a transparent rules engine to estimate head shape, then scores a curated hairstyle catalog. Users can generate an identity-preserving preview and carry the selected look into a barber booking flow.
After more than 50 barber interviews, the product moved from recommendation tool to communication layer. Resume-reported results include multiple salon adoption and a 10-minute reduction in appointment time.
System map
From ambiguity to evidence.
- 01Capture
- 02Analyze
- 03Rank
- 04Preview
- 05Book
Explainable vision
Landmarks become a reasoned recommendation.
The classifier compares face length, forehead width, cheekbone width, jaw width, and jaw angle against six head-shape profiles. Confidence reflects both geometric separation and photo quality.
Identity-preserving preview
Change the hair, not the person.
A structured image-editing prompt preserves facial identity, pose, skin tone, and lighting while applying the selected hairstyle. A local fallback keeps the product flow usable without external model access.
Commercial loop
The preview ends in a real service decision.
The booking experience carries the generated look, face-shape analysis, selected service, time, and client notes forward so the barber can prepare before the appointment.