← Back to projects
ML Product (Archive)

Virtual Fitting

Korea's first ML-based try-on service

ML/AI E-Commerce CES 2019

The Problem

Online fashion shoppers can't try clothes on. Fit uncertainty drives high return rates and low conversion. Existing solutions were either inaccurate or required specialized hardware that consumers didn't have.


The Approach

Coordinated with a Berlin-based AI research team to translate body-scanning ML models into a consumer-facing virtual try-on experience. Managed the end-to-end product: research team alignment, internal engineering, design, QA. Bridged the gap between ML capability and user experience.


Results

What I Learned

This was my first experience turning ML research into a shipped product. The Berlin team had the models working. Making them work for real consumers with real bodies and real expectations was a completely different challenge. This experience, bridging research and product, is the pattern I keep coming back to.