This post originally appeared on KDnuggets.
The letters “REV” start many words that are important in the practice of data science: revision, reveal, review, revenues, revolution, etc. Last year during our first Rev conference, data science leaders revealed many dimensions about the risks related to data handling and machine learning models in production. Quite apt since 2018 represented a worldwide reality check for the risk landscape related to data: GDPR and data privacy compliance, widespread security breaches and leaked data, cyber threats specifically targeting machine learning models, news unfolding about Cambridge Analytica, plus the growing recognition of our responsibilities for social impact through ethical data science. We reviewed the problems and solutions. We revised our ideas about priorities for data science. We looked ahead.
This year, Rev 2 explores themes about data science teams. Looking beyond problems and solutions, how do we manage the team within that complex landscape? A decade has passed since industry first began to embrace the practice of data science. We have great examples about learning data science, and how people can upskill to join industry teams – as those inspiring “Data 8” courses at UC Berkeley with thousands of students illustrate. While acquiring skills to become an individual contributor is incredibly valuable and so much in demand, the practice of leading data science teams in enterprise represents an entirely different matter.
For example, the industry’s been recognizing how machine learning compels entirely new process. Simply reapplying software engineering process, declaring an organization to be “Agile”, that’s a recipe for disaster – as last year’s themes about risk should fix firmly at top-of-mind.
Where can we look for guidance about data science leadership and emerging process? First take a close look at the successful teams. What are their best practices? Which have been their difficult lessons to learn? How can we build atop their insights and examples?
Rev 2 will feature precisely that. Come to New York City on May 23–24 to learn from data science teams and leaders at Nike, Netflix, Slack, Stitch Fix, Domino Data Lab, Microsoft, Dell, Red Hat, Google, Turner Broadcasting, Humana, Workday, Lloyds Banking, BNP Paribas Cardif, and many others about topics like:
I like to think of this year’s focus as “What can teams learn from each other?”
Headlining the Rev 2 keynotes, we’ll feature Nobel laureate Daniel Kahneman, author of Thinking, Fast and Slow. Consider: Why do we need models? Where are companies failing to use models appropriately, and why? What are some practical steps data science organizations should take to apply models successfully in their business? Much of Dr. Kahneman’s life and work has been devoted to the science of human decision making. In this era of AI applications, unbundling the process of decision-making represents a key challenge in enterprise, and Dr. Kahneman brings crucial insights.
Tom Kornegay at Nike, Michelle Ufford at Netflix, and Josh Wills at Slack – all three keynote speakers, and I’ve had the privilege of talking with their teams, hearing about the challenges they’ve faced and how they met those for data science success stories in enterprise.
Two other of my favorite speakers joining Rev 2: Maryam Jahanshahi, a research scientist at TapRecruit, presenting about how they leverage data science to help changing our thinking about how to hire data scientists, particularly with regard to fostering both expertise and inclusion. Also, Pete Skomoroch, an alumni of the groundbreaking data science at LinkedIn led by DJ Patil – Pete founded an AI company called SkipFlag focused on enterprise solutions, acquired a year ago by Workday. That said, Pete has seen the full range of data science practices from start-up to enterprise, and he’ll be presenting about Product Management for AI. In a set of recent industry surveys which Ben Lorica and I ran about AI adoption in enterprise, nearly 1 in 5 firms cited “Difficulties in identifying appropriate business use cases” as their main bottleneck for deploying machine learning in production. In other words, after you’ve hired a small army of data scientists and data engineers, there’s still a widespread talent gap for product management in AI. I’m looking forward to Maryam’s insights about hiring, Pete’s insights about product, and the whole range of Rev 2 speakers about how successful data sciences manage their practices.
Please join us at Rev 2 in NYC, May 23–24 and join that dialog.