aid: truefoundry-vocabulary
name: TrueFoundry Vocabulary
description: >-
  Domain vocabulary for TrueFoundry's enterprise AI and MLOps platform, covering
  AI gateway concepts, model serving, agent orchestration, MCP, and Kubernetes-native
  AI infrastructure terminology.
created: '2026-05-03'
modified: '2026-05-03'
tags:
  - AI Platform
  - MLOps
  - LLM
  - Kubernetes

terms:
  - term: AI Gateway
    definition: >-
      TrueFoundry's unified proxy layer that routes requests to 1000+ language
      models across 30+ providers through a single OpenAI-compatible endpoint
      with centralized auth, rate limiting, and observability.
    tags:
      - AI Gateway
      - Core Concepts

  - term: MCP Gateway
    definition: >-
      TrueFoundry's centralized registry and proxy for MCP (Model Context Protocol)
      servers. Provides authentication, access control, schema validation, and
      tool orchestration across multiple MCP servers.
    tags:
      - MCP
      - AI Gateway

  - term: Control Plane URL
    definition: >-
      The base URL for a TrueFoundry deployment. Defaults to app.truefoundry.com
      for the SaaS offering, or a custom domain for self-hosted/VPC deployments.
    tags:
      - Infrastructure

  - term: Provider Account
    definition: >-
      A configured connection to an LLM provider (OpenAI, Anthropic, Google,
      AWS Bedrock, etc.) within TrueFoundry. Each provider account has a
      display name used as the model prefix.
    tags:
      - AI Gateway
      - Core Concepts

  - term: Model Registry
    definition: >-
      TrueFoundry's versioned repository for storing and managing machine
      learning models backed by cloud storage (S3, GCS, Azure Blob, Minio).
    tags:
      - MLOps
      - Model Management

  - term: Deployment
    definition: >-
      A running instance of a service, job, or model on TrueFoundry's
      Kubernetes-based compute infrastructure. Deployments can autoscale
      including scale-to-zero.
    tags:
      - MLOps
      - Infrastructure

  - term: Service
    definition: >-
      A long-running container deployment on TrueFoundry, typically exposing
      an HTTP endpoint. Used for model serving, REST APIs, and web applications.
    tags:
      - MLOps

  - term: Job
    definition: >-
      A batch processing task that runs to completion. Used for training runs,
      data processing pipelines, and evaluation jobs on TrueFoundry.
    tags:
      - MLOps

  - term: Fine-Tuning
    definition: >-
      The process of adapting a pre-trained foundation model to a specific
      task or domain using custom training data. TrueFoundry supports
      fine-tuning workflows on Kubernetes compute.
    tags:
      - Machine Learning
      - Model Training

  - term: RAG
    definition: >-
      Retrieval-Augmented Generation — an AI architecture that combines
      information retrieval (embeddings + vector search) with LLM generation
      to produce grounded, factual responses.
    abbreviation: RAG
    tags:
      - AI Patterns
      - Embeddings

  - term: MCP
    definition: >-
      Model Context Protocol — an open protocol for connecting AI models to
      external tools, data sources, and services. TrueFoundry's MCP Gateway
      provides centralized management of MCP servers.
    abbreviation: MCP
    tags:
      - MCP
      - AI Agents

  - term: vLLM
    definition: >-
      An open-source high-throughput LLM inference engine used by TrueFoundry
      for serving large language models efficiently on GPU infrastructure.
    tags:
      - Model Serving
      - Infrastructure

  - term: Workspace
    definition: >-
      An isolated organizational unit in TrueFoundry for grouping deployments,
      models, and resources. Corresponds to a Kubernetes namespace.
    tags:
      - Infrastructure
      - Multi-Tenancy

  - term: Budget Limiting
    definition: >-
      TrueFoundry AI Gateway's cost control feature that enforces spending
      limits for users, teams, and models to prevent cost overruns.
    tags:
      - AI Gateway
      - Cost Management