Description
In this talk Elena will describe challenges that arise when building machine learning models that should run in production to personalize user's search. She will present biases that can be hidden in your observational data and can destroy your model completely and will explain several options on how you can correct for them.