Scale By The Bay 2019: Rob Munro Interview

Scale By The Bay 2019: Rob Munro Interview

Recording: Scale By The Bay 2019: Rob Munro Interview

[Music] hey I'm Robert Monroe I'm currently author of human-in-the-loop machine learning [Music] so I've been speaking at the by the bay conferences since I think the start recovered them for many years and really enjoyed them and it's been wonderful to see them grow the most important part of my talk was to let people know that there are many different ways that they can interpret their machine learning models I think a lot of people come out of college or maybe online programs knowing how to build machine learning algorithms but not realizing that there's maybe a dozen different ways that you can interpret the output of that model and what that means is that a lot of a lot of data scientists will reinvent at different companies some already well-known ways of sampling the right items so over the next year I'll be finishing publication of my book with which I'm really enjoying and for the first time it'll be one book contained in the methods that you need to sample from your machine learning models to know which unlabeled items are the best for human review do quality control over those items and then think about the right ways to update your model and all together this is something that exists in probably the majority of machine learning companies out there and so having a standard set of guidelines and best practices should really help a lot of people figure out their data management processes for machine learning [Music] what I really like about scale by the bay is that it combines people work in in DevOps and and scaling software with people who were scaling machine learning increasingly these two communities which used to be really kind of at one end of like network computing the other end of science having to work with each other when we're trying to ship and then run and distribute machine learning algorithms at scale so I think it's really nice about to bring those communities together and share experiences [Music]