Mab2Rec: A Modular Approach to Building Bandit-based Recommenders


Building recommendation models and deploying recommender systems require non-trivial effort and investment. Training recommender algorithms require sophisticated tools and expertise in feature engineering, model selection, and evaluation, while deploying large-scale recommender systems require sophisticated engineering effort.

In this talk, we present Mab2Rec, an open-source library for building bandit-based recommender systems developed by the AI Center of Excellence at Fidelity Investments. Our approach takes advantage of modular system design that tightly integrates yet maintains the independence of individual components, thus satisfying the two of the most important aspects of industrial applications, generality and specificity. This provides a powerful and scalable framework for building and deploying recommender applications, while also allowing individual components to be re-used beyond recommender systems.

Rather than creating a black-box, single-click solution to building recommender systems, Mab2Rec embraces the creativity of data scientists and provides them the tools and the suitable paradigm with several options to utilize rich types of data, train and select from many available models, and evaluate the quality of the final solution. We outline each component of the library and provide practical examples in the context of recommender systems.

Finally, we highlight our experience in scaling the knowledge, adoption, and deployment of recommender systems. Given the cross-functional nature of recommender systems we discuss the importance of maintaining a lean interface between collaborating teams as well as the need for creating a community of users that can contribute to the toolchain.

By sharing industrial-strength open-source software with the community along with practical considerations for building and deploying business-relevant applications of recommender systems, we support data science and machine learning practitioners in their own efforts to build similar systems.


Bernard is a Director in the AI Center of Excellence at Fidelity Investments working on personalization and recommender systems. His work is primarily concentrated in recommender systems and optimization, and he regularly presents on these topics, most recently at IJCAI'21 and CPAIOR'21 conferences. He is the lead developer of the open-source libraries Selective, MABWiser and Mab2Rec. He holds a MS in Computational Science and Engineering from Harvard University.

Open Data Science




Open Data Science
One Broadway
Cambridge, MA 02142

Privacy Settings
We use cookies to enhance your experience while using our website. If you are using our Services via a browser you can restrict, block or remove cookies through your web browser settings. We also use content and scripts from third parties that may use tracking technologies. You can selectively provide your consent below to allow such third party embeds. For complete information about the cookies we use, data we collect and how we process them, please check our Privacy Policy
Consent to display content from - Youtube
Consent to display content from - Vimeo
Google Maps
Consent to display content from - Google