100 Machine Learning Models running live: The Booking.com approach
Booking.com, the world’s largest online travel agent applies machine learning to improve customers experience at every step of their journey. We successfully design, build, and deploy hundreds of models that today serve predictions for hundreds of millions of users every day. Supporting Machine Learning Models at scale involves many challenges, one of them is how to ship models to production reliably, as fast as possible and accommodating a large variety of model types, invocation modes, libraries, data sources, monitoring approaches, etc. In Booking.com we believe that diversity gives us strength so we created a system to support as many ways of working as possible. In this talk we share how we get our models out of the training process to live running, by showing patterns, good practices, challenges, and practical examples.