Mellerikat
Mellerikat
Mellerikat is a tool designed for data scientists, data engineers, and AI operators aiming to automate tasks using artificial intelligence.
Traditional AI model development has involved significant time and cost due to complex compatibility issues, performance degradation from shifts in data distribution, and the need for retraining and comparison. Additionally, individuals without specialized knowledge in AI have faced considerable challenges in developing and applying models. Mellerikat addresses these issues by enabling users to develop and deploy models more easily and conveniently, supporting real-world applications.
Mellerikat supports the entire AI development process, including model creation, deployment to operational environments, integration of data pipelines for inference, performance monitoring, root cause analysis, and the creation and deployment of new models.
This manual introduces the components of Mellerikat and explains how to use them, offering a comprehensive guide for users who are new to the platform.
This manual consists of the following core features.
Key Features
ALO
AI Learning Organizer (ALO) is a specialized framework for improving the efficiency and simplification of AI model development and deployment.
Key features include converting existing AI modeling code into a format compatible with AI Conductor, and building environment information that ensures compatibility across different infrastructures.
AI Contents
AI Contents is a curated collection of pre-built AI models designed for general-purpose use by data scientists.
Key features include providing core frameworks for AI models, supporting combinations and variations of various model architectures, and offering methods for memory optimization and efficient training/inference/preprocessing.
AI Conductor
AI Conductor is a web service for developing, registering, and managing AI models within IT service operation platforms.
Key features include integration and management of various AI models, support for model optimization and training based on user data, and productivity enhancement through linkage with deployment environments.
Edge Conductor
Edge Conductor is a system that streamlines model training, deployment, and operation in on-site environments where AI models are deployed.
Key features include managing versions by configuring multiple parameters and environment settings for a single model, requesting model training via AI Conductor, and tracking training history through model versions and datasets.
Edge App
Edge App is an application installed on various devices that supports on-site operations including AI-based inference.
Key features include performing inference using AI models, collecting and validating field data, detecting and handling model errors, and transmitting collected data to Edge Conductor.