01
AI shopping companion for fashion e-commerce
Sipoza · Co-Founder & CTO · 9 months
The challenge
Fashion is one of the hardest categories to sell online. Shoppers can't try items on, so they often order several sizes and return most of them. Returns cost retailers margin and logistics effort, and they carry a sustainability cost too. Mid-market brands feel this most, as they rarely have in-house AI teams to build personalized advice themselves.
The approach
Sipoza brings the in-store advisor experience into the online shop: a white-label AI companion that recommends the right size and suggests styles that match each customer's taste.
- 01
Size prediction
Machine learning models that estimate the best fit from product data and customer input.
- 02
Style personalization
Multimodal AI (CLIP) that matches items to a customer's visual preferences.
- 03
Retailer integration
A Shopify app that lets brands add the companion to their store with minimal effort.
- 04
Platform
End-to-end architecture on Google Cloud, running multiple AI models in production.
The outcome
Sipoza went from first prototype to live pilots with fashion retailers, with 200+ shoppers already using the companion.
Supported by
Berlin Startup Stipendium · HEC Paris Incubator · Potsdam Impact Labs · ESCP Blue Factory · Encourage Ventures
My role
Co-founder, CTO and sole technical co-founder. I own the product architecture, the AI pipelines and the infrastructure, and I lead engineering from MVP to production.