Abstract: Every year, our Data Science community at Liberty Mutual comes together in an incredible way – to collectively solve the organization’s toughest problems using Artificial Intelligence, Machine Learning, and Data Science. Like Kaggle or the Netflix Prize, every year we curate a particular opportunity to solve with AI, structure a challenge, and post a bounty for the scientist or team who can provide the best solution. In the past, we have achieved what was previously unimaginable. We improved our core pricing models with advanced techniques. We defined the ‘style’ of drivers using sensor data. We used photos and written notes to predict the outcome of a claim. But more importantly: we got better at data science!
By embracing new challenges cooperatively, everyone learns – whether how to train a bleeding edge AI model, or more simply, how to best practice clean and repeatable science through code. We design the challenge each year to deliver value to a core project, as well as to practice, refine, establish standards and training material, and more broadly invest in our common DS toolbox. In this talk, we will discuss what we have learned about how to uplift an enterprise with a collective call-to-arms, and how you can bring grass-roots innovation to your organization as well.
Bio: Scott Gorlin, Ph.D., is the Senior Director of the Science team in the Office of Data Science, a centralized group within Liberty Mutual focused on catalyzing applications of AI/ML, designing core tools and platforms to accelerate DS and MLOps, and broadly setting best practices for applied science across the organization.