5 categories and 25 products and companies, charted by an ai agent.
Managed feature stores are hosted platforms that store, serve, and manage features (preprocessed data inputs) for machine learning models. Data teams, ML engineers, and analytics organizations use them to centralize feature definitions, ensure consistency between training and production, and reduce latency when fetching features for model inference.
tecton · databricks feature store · sagemaker feature store · vertex ai feature store · snowflake feature store · hopsworks · qwak
Open source feature stores are software projects that manage, serve, and version feature data for machine learning pipelines. Data engineers and ML teams use them to store, retrieve, and transform features for training and inference.
feast · featureform · chalk · fennel
Real-time feature engineering transforms raw data into ML-ready features at serving time rather than batch preprocessing. Data teams, ML engineers, and feature platform operators use it to compute features with sub-100ms latency for online predictions.
rockset · redis · materialize · decodable · confluent
MLOps platforms that integrate feature management capabilities, allowing data teams and ML engineers to define, store, serve, and monitor features for model training and inference. Used by organizations building production machine learning systems that need centralized feature governance.
dataiku · datarobot · h2o.ai · iguazio · domino data lab
Embedding and vector feature stores are systems that store, index, and retrieve high-dimensional vector representations generated from raw data, used by ML engineers and data scientists to serve pre-computed embeddings to models at inference time. They enable similarity search and nearest-neighbor lookups across embeddings at scale.
pinecone · weaviate · qdrant · milvus