By Mac Steele, Director of Product, Domino on June 01, 2018 in Perspective
This week we held our first ever Rev summit to gather more than 300 data science leaders in San Francisco for two fantastic days. Speakers and attendees discussed the strategic direction of the data science industry and shared tactical tips among fellow leaders. We, Domino, hosted Rev because we heard and felt a gap in the market for a forum where leaders could come together and take away something substantive and practical. Something that they can put to work on Monday morning. The outcome exceeded our best expectations. A few highlights we wanted to share for those who couldn’t make it in person.
Our CEO, Nick Elprin, unveiled the Domino vision for the data science space. In his keynote, he discussed how the community of data science leaders has a once-in-a-lifetime opportunity to help build a new organizational capability and unlock the full potential of data science. He detailed a new framework for this capability, termed Model Management, and shared how we are bringing this to life with a demo of the future of the Domino data science platform. If you’re curious to learn more, please check out this paper which introduces Model Management in more detail.
We heard thought-provoking keynotes from Cathy O’Neil and Nate Silver.
Cathy, author of Weapons of Math Destruction, spoke about the importance of taking a broad perspective on what constitutes a “good” model given all the different stakeholders involved. She suggested teams use a tool called an ethical matrix before blindly diving straight into model building, particularly as models come to drive our interactions with most bureaucracies.
Nate, author of The Signal and the Noise, discussed the importance of probabilistic thinking and how the explosion of models fueled by big data has created a number of problems and opportunities for society. He highlighted the need for rapid iteration, citing the dramatic improvement in weather forecasting as meteorologists constantly learn and adjust from past mistakes.
Leaders from model-driven companies like Stitch Fix, Google, Uber, Airbnb, UnitedHealth Group, JPMorgan Chase and others shared their own perspectives. Nathan Siemers of Bristol-Myers Squibb detailed how his team uses models to discover genetic biomarkers to improve cancer treatment. Jacob Grotta from Moody’s Analytics described the storied history of models in the banking industry and how they have developed a rigorous system for managing those models from which other industries can learn. He noted that all data scientists like to think of themselves as artists, and he said at Moody’s the true art is “Auditable, Repeatable, and Transparent.”
There were a number of panels and an interactive workshop on how to best manage data science organizations. Leaders engaged in a healthy debate about which roles to hire for, in what order, and who Data Science should report to. The importance of collaboration and the goal of reproducibility were recurring themes across sessions. Many attendees described a feeling relief that the challenges they face are not unique to them and there was a shared excitement to be shaping the best practices in our still nascent industry.
Overall, it was an invigorating two days and we are humbled by the willingness of the speakers and the attendees who shared their insights to help advance the fledgling practice of data science management. We came away inspired as data science leaders across industries continue to build model-driven companies.
Stay tuned for more on Rev 2019 and, in the meantime, keep an eye out for a Data Science Pop-up in a city near you in the second half of 2018.