Skip to main content
Version: Next

Terminology

Updated 2025.03.19

This document serves as a manual for Mellerikat, an MLOps technology that may be difficult to understand for those unfamiliar with the field. It includes numerous technical and domain-specific terms that are essential for users to understand and use the system effectively. Familiarizing yourself with these terms in advance will help you read and apply the manual more efficiently. We recommend reviewing this terminology section before proceeding, to ensure a smoother and more productive experience.


Mellerikat-specific Terminology​

  • AI Solution: AI model code developed to solve a specific problem tailored to a business case or dataset.
  • Instance: An individual computing resource running on a cloud or server environment.
  • Stream: A pipeline that manages the entire workflow from data processing to model training and deployment.
  • Metadata: Data that describes or provides information about other data.
  • Edge: A computing environment where data generation or processing occurs at the outermost part of the network, often on devices.
  • MLOps: A comprehensive framework that applies DevOps principles to machine learning for the development, deployment, operation, and management of models.

AI Terminology​

  • Asset: Valuable resources used in machine learning such as datasets, models, features, and code.
  • Pipeline: An automated sequence of steps for data processing, feature extraction, model training, and evaluation in machine learning.
  • Deployment: The process of installing a trained model onto an Edge App to enable real-time inference.
  • Re-labeling: The process of correcting or modifying labels in a dataset within the context of machine learning and data science.
  • Feature Engineering: The process of creating, selecting, and transforming data features to improve machine learning model performance.
  • Artifact: Output generated during the software development process, such as models, reports, and training logs.

MLOps and Cloud Infrastructure Terminology​

  • Protocol: A set of rules or standards for data communication between computers.
  • Workflow: A defined sequence of steps or procedures for accomplishing a specific task.
  • Access Token: A temporary key used for authentication and authorization in a system.
  • Config: A file or data that defines the configuration settings of a system.
  • On-premise: Software or hardware deployed and run in a physical, in-house environment rather than in the cloud.
  • Workspace: A configured environment for performing specific tasks.
  • Dashboard: A user interface that visually represents data to provide an at-a-glance overview of system status.
  • Bucket: A logical container used for storing data in a cloud storage service.
  • API: An interface that enables interaction between different software applications.
  • Migration: The process of moving systems or data from one environment to another.

  • WSL: Windows Subsystem for Linux, a feature that allows the native execution of the Linux kernel on Windows.
  • Docker: An open-source platform for containerizing and deploying applications.
  • Container: A packaged unit that allows an application and its dependencies to run in an isolated environment.
  • Kubernetes: An open-source system for automating the deployment, scaling, and management of containerized applications.
  • EKS Cluster: A managed Kubernetes cluster provided by AWS (Amazon Web Services).
  • Nodegroup: A group of nodes within an EKS cluster.
  • Kubeflow: An open-source project for running machine learning workflows on Kubernetes.
  • AWS: Amazon Web Services, a cloud computing services platform.