Speaker "Davor Bonaci" Details Back



Time for quality ML from event-based data


Organizations have been trying to solve problems with ML for years and still it takes 7 - 18 months on average to go from idea to ML model in production. Those that make it are often forgotten and are underperforming or broken for years. But things are changing. Data platforms are maturing, and where success was difficult, it is getting within reach of many organizations.
We are now entering a new phase, however. There has been so much investment in the space over the years. We can now build and operate several classes of machine learning, predictably and reliably, and it is within reach of many of us, not just the Big Tech with their unlimited resources. Now is the time to think again about how we can solve the underlying business problems. 
Come hear about some top level trends. Be wary about building / rebuilding ML platforms. The costs and complexities are very high.


Davor is a co-founder and Chief Executive Officer at Kaskada, a machine learning startup. Kaskada transforms data science with the first system for iterative time-based feature engineering, significantly improving quality and availability of machine learning to many more companies. Previously, Davor served as the chair of the Apache Beam Project Management Committee and an engineer in Google Cloud since its early days and the inception of Cloud Dataflow. Davor holds a Master's degree from the University of Washington.