Senior Applied AI/ML Scientist (Listing Quality)

Join Faire, a company revolutionizing the wholesale industry through machine learning and data insights. As a Senior Applied AI/ML Scientist on the Listing Quality team, you will own the modeling and measurement for the content that powers Faire's catalog. You will work primarily with unstructured data using multi-modal deep learning and LLMs, and drive projects end-to-end from framing through production and measurement. You will collaborate with product, engineering, design, and analytics teams to turn models into shipped products and business impact.

- Posséder des projets d'apprentissage automatique appliqué de bout en bout, de la définition du problème à la mesure de l'impact. - Utiliser l'apprentissage profond multimodal et les modèles de langage de grande taille pour comprendre le contenu des annonces et détecter les problèmes de qualité. - Améliorer les images de produits et le texte des annonces, et mesurer l'effet en aval sur la découverte et la conversion.

- 3+ years of industry experience using machine learning to solve real-world problems - Experience with relevant technical methods (deep learning and LLMs, computer vision, information extraction, entity resolution, ranking, and/or experimentation and causal inference) - The ability to drive a project end-to-end and lead model development with limited supervision - An excitement and willingness to learn new tools and techniques - Strong communication skills and the ability to work in a highly cross-functional team - Strong programming skills - Experience with relevant business problems (e-commerce, marketplaces, catalog and content quality, search, or personalization) - Previous experience with catalog quality, product attribute extraction, computer vision for e-commerce imagery, or search and discovery for a two-sided platform - Highly recommended: Master's or PhD in Computer Science, Statistics, or related STEM fields - Experience building and validating LLM evaluation pipelines, including prompt iteration against labeled data and human-in-the-loop workflows