Government

[KITECH] Building an Integrated Platform for an HPC-based AI R&D Environment

To support the digital transformation and value creation of the engineering industry, the Korea Institute of Industrial Technology (KITECH) built an AI analytics support platform based on high-performance computing infrastructure. The platform provides an integrated R&D environment where industry, academic, and research users can carry out everything from big data analytics to AI model development, deployment, and asset sharing.

CheetahMLOps

An AI R&D Environment for Leveraging Engineering Big Data

KITECH aimed to build a foundation for boosting the competitiveness of the engineering industry by applying digital technologies such as AI and big data. To do this, it needed an integrated environment that efficiently operates large-scale HPC infrastructure while allowing industry, academic, and research users to use big data for AI model development and validation.

Securing HPC-based AI R&D and Asset Utilization

The key challenge was to reliably operate high-performance computing resources while letting many users perform big data analytics and AI R&D in independent environments, and share and reuse their results.

  • HPC infrastructure operations: A framework to efficiently manage and operate 48 high-performance A100 GPUs

  • Multi-user research environment: A big data analytics and AI R&D environment for many concurrent industry, academic, and research users

  • ML pipelines and asset management: A framework spanning data management, model development and deployment, and sharing of research and training assets

Building an Analytics Support Platform Covering HPC Operations and AI R&D

With Cheetah, we virtualized high-performance infrastructure for efficient operation and provided big data analytics and AI development environments, one-click deployment, and research and training asset management.

  • Virtualized HPC resource management: Efficient operation and allocation of high-performance resources on Kubernetes

  • AI development and deployment process: An R&D environment covering data management, model development, and one-click deployment

  • Management and sharing of diverse assets: Integrated management and sharing of research and training assets such as Jupyter Notebooks, PDF, MP4, and PPT files

More Efficient AI Research and Better Infrastructure Utilization

KITECH now operates its 48 A100 GPUs more efficiently and can reliably provide users with as many resources as they need. Users can quickly set up independent AI development environments and manage and share diverse research and training assets on one platform, making analysis, development, and hands-on work more efficient.

  • More efficient infrastructure operations: Allocates high-performance GPUs as needed, reducing waste and improving efficiency

  • Faster R&D environment setup: Users quickly set up independent AI development environments for analysis and development

  • Better asset use and collaboration: Integrated management and sharing of AI assets boosts their use in research, training, and practice

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

Have more questions?

Contact us anytime to learn more about AIFRICA solutions.

Contact Us