Zurich Insurance on Open Source Machine Learning
- H2O-3
Overview
Connor Jensen, analytics program director and interim lead data scientist at Zurich North America, explains why the insurer made open source machine learning a core strategic investment. He covers why scalable, distributed, in memory analytics matters for a business built on statistics and probability, why open source flexibility mattered more than being locked into a single vendor's toolset, and how tools like H2O Flow helped the team explain models and build end user trust. He also touches on how the open source investment helped with recruiting in a tight data science talent market.
Key takeaways
- Why Zurich North America made predictive analytics and open source tooling a key strategic investment
- Why distributed learning and in memory processing mattered as the company looked to scale analytics across growing data volumes
- Why Zurich prioritized open source over being tied to a single vendor's toolset, valuing the flexibility to mix and match tools
- How tools like H2O Flow helped the team tell the story behind a model's data and predictions, building end user confidence
- How open source let the team test new technology without a large upfront bet, scaling what worked and moving on from what didn't
- How the open source approach helped with recruiting in a data science talent market moving quickly
Start owning your AI