Subject: Seeking guidance on AI implementation on IBM Z & LinuxONE

Hello IBM Community members,

I'm reaching out to seek guidance and recommendations from this community regarding the implementation of AI solutions on IBM Z & LinuxONE.

I'm particularly interested in learning about best practices, available tools, and any experiences you may have had in this area.

Thank you in advance for your insights and support.

Hi Martin. 

If you are in EMEA, tell your IBM contact to engage me, it will be faster and more effective than discussing here. I can help getting into the correct path and addressing the specific needs. 

A good place to start reading and following all links in there is the AI on IBM Z 101

Regards, 

------------------------------
Diego Cardalliaguet
IBM Data&AI zStack Leader for EMEA
IBM
------------------------------

Implementing AI on IBM Z and LinuxONE is about leveraging their inherent power and security. The main best practice is Data Proximity-you want to run your AI models right where your sensitive transactional data already resides to minimize latency and risk. The good news is these platforms fully support the standard open-source AI ecosystem, which includes Python environments and major frameworks like TensorFlow and PyTorch, all optimized for the $s390x$ architecture. For development, you'll rely on tools like Anaconda and Jupyter Notebooks. When it comes time to deploy and manage these models, you'll want to containerize them using Docker or Podman, ideally orchestrated with OpenShift, and manage the lifecycle (MLOps) using platforms like IBM Watson Machine Learning or MLflow. Simply put, you get the performance and security of the mainframe with the flexibility of open-source AI tools, making it perfect for high-speed, secure tasks like real-time fraud detection.

------------------------------
john wick
------------------------------