Deploying distributed AI and machine learning in financial services
summary
Deploys H2O stack on Hadoop Spark and Kubernetes to train distributed models and export portable binaries, claiming H2O Core algorithms scale to petabytes and Driverless AI performs genetic feature evolution over ~40–50 iterations with hundreds of models and thousands of candidate features. Outputs a binary scoring artifact and a 30–50 page model report, runs on the MOJO runtime, and ships to Kubernetes MLOps, AWS Lambda, Kafka, databases or REST with monitoring, drift detection and champion/challenger.
AI-written summary · about
chapters (11)
- 0:00Introduction to talk
- 1:25Data foundation for AI
- 3:31H2O platform overview
- 5:14Distributed ML pipeline
- 9:46Enterprise ML challenges
- 12:12Automated ML evolution
- 15:55Driverless AI features
- 19:47Model interpretability
- 21:38Model deployment options
- 24:49Automation levels summary
- 25:38Q&A session
related talks
44mScalable Automatic Machine Learning with H2O
50mScalable Data Science & Deep Learning with H2O
30mScalable Automatic Machine Learning with H2O
33mProductionizing H2O Models with Apache Spark
27mDeveloping and Deploying Models at Scale — without Spark!
01h 18mIntroduction to H20 AutoML with Python
40mHow to leverage the Apache Kafka Ecosystem to productionize Machine Learning
01h 10mIntroduction to Machine Learning with H2O and Python