Hopsworks

Hopsworks is an AI lakehouse and machine learning platform built around a feature store. It covers the full ML lifecycle in one system: a central repository for feature data with sub millisecond online retrieval, a model registry, experiment tracking, model serving and deployment pipelines, and monitoring. The feature store is powered by RonDB, an open source key value database, and lets teams reuse features across models and keep training and production data consistent. The lakehouse works directly with Delta, Iceberg and Hudi tables and supports Spark, Flink, Pandas and DuckDB. For compute, it provides GPU scheduling and quota management, training with Ray and serving with KServe and vLLM. Deployment options include managed SaaS, Kubernetes on AWS, Google Cloud, Azure and OVHcloud, and air gapped or on premises installs for sovereign AI requirements. The vendor also lists coding agent support inside the platform. It is aimed at data science and ML engineering teams that run real time and batch production systems, such as fraud detection, personalization and credit decisioning. The Free plan costs $0 with no credit card, one project, the feature store and model registry, and community support. The SaaS plan is pay as you go with usage based pricing, unlimited projects, model serving and a platform SLA. Enterprise is custom priced and adds on premises and air gapped deployment, dedicated support and custom integrations.