Hopsworks
Hopsworks AB develops an open, Kubernetes-native AI Lakehouse platform that integrates feature engineering, model training, real-time inference and vector search within a single stack. The platform can be deployed on-premises, in air-gapped environments or in any cloud, supporting requirements for data sovereignty and security.Originating from research at KTH Royal Institute of Technology, Hopsworks builds on more than a decade of experience in translating peer-reviewed systems, including RonDB and HopsFS, into production-ready infrastructure used by research organisations and regulated industries across Europe.
The platform is designed to meet the needs of large-scale scientific environments, offering support for petabyte-scale data management, reproducibility, data lineage and governance, as well as GPU-aware workloads. It supports widely used frameworks such as Python, Spark and Flink, along with emerging AI and large language model applications.
Core Competencies
- Platform & Architecture: AI Lakehouse, Feature Store, MLOps Platform, Kubernetes-native, Sovereign AI, On-premises Deployment, Air-gapped Environments, Hybrid Cloud, Open Source- Data Infrastructure: RonDB, HopsFS, Delta Lake, Petabyte-scale Storage, Data Lineage, Data Governance, Time-travel, Reproducibility
- Compute & Orchestration: GPU Scheduling, Ray, Apache Spark, Apache Flink, Apache Kafka
- Distributed Training & AI Workloads: Real-time Inference, Online and Offline Feature Serving, Vector Database, Retrieval-Augmented Generation (RAG), LLM Fine-tuning, Agentic AI, Model Registry, Model Monitoring
Industry Sectors
- Big Science & Research Infrastructure- High-Performance Computing (HPC)
- Artificial Intelligence & Machine Learning
- Financial Services
- Telecommunications
- Automotive
- E-commerce & Retail
- Healthcare & Life Sciences
- Public Sector & Sovereign AI
- Energy
References
- Karolinska Institutet: One of the world’s leading medical research universities. Hopsworks supports large-scale data and machine learning workflows for academic research, with strong relevance to Big Science requirements for reproducibility and data governance.- Human Exposome Assessment Platform (HEAP): EU-funded open research platform studying the impact of the exposome on human health. Selected Hopsworks for its ability to manage large volumes of heterogeneous scientific data in a reproducible manner.
- Saab: Uses Hopsworks for mission-critical machine learning in regulated and sovereignty-sensitive environments, relevant to air-gapped and on-premises deployments in Big Science.
- Ericsson: large-scale machine learning infrastructure supporting network optimisation and predictive maintenance in production environments.