Seldon
Kubernetes-based platform for deploying, scaling, and monitoring machine learning models in production.
📍 United Kingdom 🇬🇧, London
Product overview
Seldon provides an open-source-rooted platform for deploying and managing machine learning models at scale on Kubernetes. Organizations use it to serve models in production, run A/B tests, detect data drift, and monitor performance across hundreds or thousands of deployed models. The product line has three tiers. MLServer is an open-source inference server (Apache 2.0) that handles model serving with pre-built runtimes for common frameworks. Core 2 adds a Kubernetes-native framework for pipeline orchestration, autoscaling, shadow deployments, and streaming via Kafka. Core+ layers on commercial support with add-on modules: Alibi Explain for model interpretability, Alibi Detect for drift and outlier detection, and an LLM Module for deploying generative AI applications including RAG pipelines. Since January 2024, Core and Alibi products use a Business Source License (BSL 1.1), free for development and testing but requiring a commercial license for production use. MLServer remains fully open source. The platform is framework-agnostic and can serve any model, including open-weight European LLMs, on any Kubernetes-compatible infrastructure. Founded in London in 2014, Seldon counts Capital One, Ford, PayPal, and Covea Insurance among its clients. The company reports over 1.7 million unique models deployed across its enterprise customer base. KEY FEATURES: - MLServer open-source inference server (Apache 2.0) - Kubernetes-native model deployment with autoscaling and A/B testing - Alibi modules for drift detection and model explainability - LLM Module for generative AI and RAG deployments - On-premise or any cloud deployment on Kubernetes