H2O-3

This post describes the core H2O-3 open-source distributed in-memory machine learning platform. It details the platform's key features including leading algorithms (GBM, GLM, deep learning, XGBoost, GLRM, Word2Vec), its AutoML functionality for automating model selection and hyperparameter tuning, distributed in-memory processing for scalability and speed, and simple deployment options via POJOs and MOJOs. The post also mentions its ability to run on existing big data infrastructure (Hadoop, Spark, Kubernetes) and ingest data from various sources (HDFS, Spark, S3, Azure Data Lake).