Machine learning tools are rapidly maturing. TensorFlow gave developers an open source version of Google’s internal machine learning framework. Cloud computing provides a cost effective, accessible way of training models. Edge computing allows for low latency deployments of models.

But even if you are a kid with a laptop who has learned all the machine learning algorithms, read all of the deep learning textbooks, and figured out how to use AWS, all of the tooling and education in the world doesn’t change the fact that you still need data to build models.

This illustrates why we need data-as-a-service.

A kid with a laptop has access to infrastructure-as-a-service, platform-as-a-service, and software-as-a-service. As these tools build on each other, there has been an explosion of high-leverage software products. But the world of data sets remains crude and underdeveloped.

Think about some data sets you could take advantage of the number of emergency room patients that come into a hospital with chest pain; the size of the average coffee mug; the principal component breakdown of sidewalk concrete in San Francisco.

SafeGraph is a company that offers data sets as a service. Auren Hoffman is the CEO of SafeGraph, and he joins the show to discuss why he started building SafeGraph and how he thinks about the state of publicly accessible data.

Auren was previously on the podcast, and I always enjoy talking to him–this was a great episode and I think you will like it as well. Full disclosure: LiveRamp is a sponsor of Software Engineering Daily, LiveRamp being the company that Auren created prior to SafeGraph.

Raj Chetty economic papers

Paul Graham “Keep Your Identity Small”

Auren Hoffman on Quora

The post SafeGraph with Auren Hoffman appeared first on Software Engineering Daily.