#066

I was on support Monday and Tuesday, it was nicely uneventful.

We ran the Learning to Rank prototype for a day with 50% of production traffic, and it worked pretty well - all the metrics were moving in the right direction. We want to run a longer test to get more data, but not while almost everyone is out over the Christmas period (in case something goes wrong).

I started looking into how we can use AWS SageMaker for training and hosting the model.