OnDemand: AI for Obesity Research and Clinical Practice-Evaluating Competencies in Data Science

Video Transcription

Video Summary

The talk argues that organizations should not hire a “data scientist” without first understanding what problems they need solved and whether their culture, data infrastructure, and literacy are ready. Using military and medical analogies, the speaker explains that data science is only one part of a larger team that may also need data engineers, machine learning engineers, statisticians, software engineers, and “data translators.”<br /><br />A true data scientist should be able to apply statistics, basic math, coding, algorithm selection and modification, work with structured and unstructured data, communicate findings clearly, and understand ethics and bias. They are not expected to do everything, especially not tasks better handled by engineers or statisticians.<br /><br />Before hiring, organizations should assess their data strategy, policies, tools, training environment, and data literacy. Everyone should be able to read, write, and communicate with data. The speaker recommends trying a data science workflow on a real problem first, identifying pain points, and then hiring to fill those gaps. He emphasizes embedding data staff close to the problem, investing early in data engineering, and not expecting one person to solve all data challenges.

Asset Subtitle

By Nicholas J. Clark, PhD

Keywords

data science

hiring strategy

data literacy

data engineering

machine learning engineer

statistics

organizational readiness