Transformers for Recommender Systems in Fashion Ecommerce

In the age of online fashion retail, retailers can offer consumers a huge number of garments and accessories, resulting in an information overload for customers who are shopping for their next outfit. Recommender systems seek to alleviate this by surfacing the most relevant items by presenting users with personalized suggestions that match their style preferences. Recently, recommender systems have been through a paradigm shift toward sophisticated sequence-based transformer recommender systems. We'll detail the development and implementation of such a recommender system at the online fashion retailer ASOS.com, using NVIDIA’s Merlin and Transformers4Recs libraries.