We've worked on 100+ data science projects with dozens of organizations, from new startups to large foundations and Fortune 50 companies.

We've seen thousands of data science projects submitted to our data science competitions by practitioners and researchers.

As the field develops, it is becoming increasingly important to organize data science work so that it is easy to understand, reproduce, and build upon.

Here's what we've learned about the best practices—and the perils—for data science workflows.