On-Premise refers to a method where a company builds and operates its IT infrastructure directly in its own data center or servers, instead of using an external cloud. By adopting the On-Premise approach, companies gain the advantage of directly controlling data security and systems.

This document technically explains the main benefits expected when introducing Clevi On-Premise and the secure isolated computing environment for the Clevi-X MDLM (Massive Multimodal Language Model) model.

1. Benefits of Adopting On-Premise

Maximized Data Independence and Security

All processes from data input to analysis, service, and backup are operated independently within the company's closed network.
→ Minimizes risk of external leakage, optimized for data sovereignty and personal information protection
Sensitive data such as Fine Dataset and private datasets are isolated in separate security zones
→ Makes data access control and auditing easy.

High-Performance and High-Availability Infrastructure Design

GPU/CPU servers are clustered in the analysis and computation area
→ Optimized for large-scale AI computation, real-time data processing, and parallel analysis
HA (High Availability) structure applied
→ Enables stable operation without service interruption even in case of failure.

Flexible Data Management and Scalability

By separating the integrated database cluster and cold backup storage
→ Efficiently manages real-time and long-term stored data
Analysis, service, and data management areas are modularized, making expansion and maintenance easy
→ Can quickly respond to business changes, data growth, and new service introduction.

SVCE (Clevi Safe Virtual Compute Environment)

Enhanced Security
→ Completely isolates each task and user session to block malicious code execution, data leakage, and system intrusion risks.
Flexibility and Scalability
→ Dynamically adjusts the number of instances and resource allocations as needed, providing a customized environment for various AI workloads (text, image, code, etc.)
Operational Efficiency
→ Maximizes infrastructure utilization through automated resource management and scheduling, and supports rapid instance recreation and rollback in case of failure.

2. Secure Isolated Computing Environment for Clevi-X MDLM Model

Clevi-X MDLM (M Double LM - Massive Multimodal Language Model) is a next-generation multimodal AI model designed to support actual computing tasks by learning human computing usage environments, methods, structures, and procedures.

CLK (Clevi Linux Kernel)-based lightweight container environment is designed to make virus intrusion and malicious access as well as any operation beyond permissions impossible.

Even in cloud computing environments where multiple anonymous users connect for various purposes, SVCE (Safe Virtual Compute Environment) completely isolates each session and task, ensuring safety even against intrusion or attack attempts by malicious users.
All tasks such as internet access, computation, code execution, result verification, and file input/output are performed in a secure isolated space, ensuring the reliability of the model's output data and task execution capability.
Provides a secure environment against various threats such as hacking, data leakage, attacks using the model, and attacks targeting the model.

Clevi On-premise