Enterprise

[Celltrion] Building an AI Server and Ops Platform for Celltrion

With limited ability to use external AI models, Celltrion selected Cheetah as its AI platform to drive AX across internal research and business operations. Together, we built an environment for phased GPU expansion and company-wide AI development and operations.

CheetahMLOps

Building an AI Platform to Expand Company-wide AX and Internal Research

In an environment where the use of external AI models was restricted, Celltrion set out to build an integrated AI platform that supports in-house AI model development and GPU resource operations, in order to expand AI adoption across company-wide AX initiatives and internal research.

Establishing an Operational Foundation for Air-gapped AI Development and GPU Infrastructure Expansion

Celltrion needed to develop its own AI models in an air-gapped network, connect existing research data with GPU infrastructure, and at the same time put in place an integrated operating system that could keep up with ongoing GPU expansion.

  • Air-gapped AI development environment: A platform was needed to develop and operate in-house AI models where external AI models could not be used.

  • Connecting research data with GPU resources: An integrated research environment was needed to link large volumes of existing research data with new GPU infrastructure.

  • Scaling with GPU infrastructure growth: A stable resource management and operating system was needed to support continued GPU expansion.

An Integrated Platform for AI Development and GPU Resources

Built on the Cheetah platform, we connected existing research data with GPU infrastructure to create a single integrated environment for both AI model development and GPU resource management.

  • AI model development and operations environment: Set up an environment for developing and operating AI models within the air-gapped network using Cheetah.

  • Research data utilization environment: Connected the platform to existing large-scale research data so it can be used for AI model development.

  • Unified GPU resource management: Built an operating environment that manages GPU resources centrally on the platform and scales with future infrastructure expansion.

Improved AI Research Efficiency and GPU Resource Utilization

A standardized AI development environment and unified GPU resource management increased the use of research data and laid the groundwork for expanding AI across the company and continuously growing GPU infrastructure.

  • More efficient AI research and development: Systematized AI model development and operations on a shared development environment, improving research productivity.

  • Wider use of research data: Linked large volumes of existing research data to AI model development, expanding how existing data can be used.

  • More efficient GPU operations: Increased GPU utilization through resource management and secured an operating foundation ready for ongoing infrastructure expansion.

  • Are you curious about an enterprise AI integration platform? → Cheetah

Have more questions?

Contact us anytime to learn more about AIFRICA solutions.

Contact Us