Circler Fashion
AI Product Engineering

// OBJECTIVE

Circler needed to automatically tag thousands of pre-loved fashion items from photographs and match them to individual user preferences without manual curation.

// THE BREAKTHROUGH

We built a vision-model pipeline that reads item images against a structured fashion ontology—identifying garment type, cut, colour, condition, and style attributes—then feeds these tags into a recommendation engine.

// APPROACH

Ontology-driven image classification for consistent tagging at scale. User preference modelling to proactively match items. Auto-generated stylist notes that explain why a specific item fits a user's profile.

// OUTCOME

Automated tagging pipeline replaces manual item classification. Users receive personalised recommendations with human-readable explanations for each match.