Breakthroughs in data science, particularly in deep learning and neural networks, are opening up capabilities unimaginable only a decade ago. Computers now see, hear, and understand. We can create user experiences that are highly personalized with little prior information about our customers. But we have a problem. All these breakthroughs are dependent on reliable data. Unreliable and insufficient data causes poor and sometimes dangerous outcomes, often outcomes that are the exact opposite of our intentions. Open source provides tools necessary to detect and circumvent these data issues. Importantly, executives and technical teams both have essential roles in understanding these problems and need to communicate well with each other on a continuous basis. We’ll walk through the concepts, processes, and tools, and provide best practices for all teams in the emerging field called human-centered AI.