Seattle, Washington, United States
Specialized in machine learning, I spent 9+ years as an engineer in Bing Multimedia, building full stack solutions to model and understand users' search queries, detect and remove offensive and harmful content from our platform, and built our related searches generation system. I now lead the team of software engineers and data scientists responsible for these features, and help make Bing the most useful and safe place to search online.
Leading the Query Intelligence and Related Searches team at Bing Multimedia, applying machine learning large scale and building full stack solutions to model and understand user queries, remove offensive and harmful content from our platform, and provide useful search suggestions to help our users in their search journeys.
In the Bing Multimedia Search team I work with big data daily, building large scale machine learning algorithms, natural language processing and other data mining solutions in a distributed environment, collaborating closely with the UX team to drive user engagement. Select projects and systems built: - Bing Multimedia Related Searches: data generation, filtering and ranking, and everything else in between. - Image Recommendations: architecture and implementation of a system to recommend high quality, visually appealing and diverse images tailored to the users' specific tastes - Image Tagging: design and implementation of a pipeline generating titles for billions of images. Directly responsible for millions of monthly queries and used throughout Bing to improve the relevance of many features - Distributed Matrix Factorization: novel distributed Singular Value Decomposition algorithm capable of handling billions of training samples
Développement d'un site intranet en PHP5/SQL pour publier et visualiser des recommandations ergonomiques. Création et implémentation d'un moteur de recherche et d'un système d'autocomplétion utilisant Ajax et Mysql. Suivi d'un processus ergonomique : analyse des besoins, analyse des utilisateurs et entretiens en Anglais, tests avec des utilisateurs finaux.
Lors du Google Summer of Code 2008 : Implémentation du moteur de recherche utilisé sur le site http://support.mozilla.com - Ajout de nouvelles fonctionnalités au moteur de recherche Sphinx en C++ - Intégration dans le CMS en PHP sur lequel le site était construit