Accelerating Generative AI with HPE: From Development to Deployment

Dmitry Spodarets -

This webinar will cover:

Who is HPE in AI?

It’s important to clarify who we are and what we do, as many AI/ML professionals aren’t fully aware of our capabilities. I won’t talk about AI in general — everyone already knows about it. This message is for a technical audience, including AI startups and IT system integrators, who are already familiar with Generative AI and its use cases.

​Why HPE for Generative AI?

HPE stands out due to several key strengths:

  • ​Strong collaboration with key component vendors worldwide.
  • ​Supercomputing experience and market leadership.
  • ​Full-stack software and hardware approach, combined with services.

​These elements make us uniquely positioned to deliver comprehensive AI solutions.

​Examples of HPE Hardware and Software Solutions

While customers can buy individual components from us, like they might from Supermicro, Lenovo, or Dell, that’s just a small part of the value we offer. HPE helps build and deliver complex, integrated solutions that scale from small setups to large deployments, customized to the customer’s needs for on-premise environments.

  • ​GenAI Starter/Development Kit. This solution is designed for AI development and experimentation on-premise. It includes a single server bundled with GPUs, software, and consulting services. Everything is integrated, installed, and delivered directly to the customer, ready for immediate use. Additional services ensure the system is properly set up for the customer’s needs. The kit also includes RAG (Retrieval-Augmented Generation) and code generation pipelines, all configured for rapid deployment.
  • ​PCAI – Private Cloud AI Enterprise. A turnkey enterprise-grade AI cluster, pre-configured with several servers, various software solutions, and bundled services. It’s designed for immediate use, leveraging NVIDIA and HPE hardware. This solution is perfect for businesses looking for a ready-to-use, integrated AI environment.
  • ​Full Scale AI Factory. A large-scale solution built for enterprises or cloud providers. It comes with distinct functionality for managing security, resource allocation, billing, and external integrations. This system supports complex AI operations with the ability to manage large-scale deployments effectively.
  • ​HPE Machine Learning Data Management (MLDM in short or Pachyderm as opensource version): provides version-controlled data pipelines for managing large-scale datasets in machine learning workflows. It ensures reproducibility and traceability by tracking changes to data and automating complex data transformations. With its scalable architecture, MLDM can handle diverse data sources and processing workloads, making it ideal for iterative model training and continuous data updates.
  • HPE Machine Learning Development Environment (MLDE in short or Determined AI as opensource version): MLDE offers a platform for building, training, and optimizing ML models at scale. It simplifies model experimentation through features like hyperparameter tuning, experiment tracking, and model parallelism. Designed for collaboration and ease of use, Determined AI accelerates training on distributed infrastructure, ensuring faster iterations and more efficient use of resources for AI projects.
  • HPE Machine Learning Inference Server (MLIS in short): MLIS provides a robust, scalable solution for deploying trained models into production environments. It supports a variety of machine learning frameworks and is optimized for high-throughput, low-latency inference tasks. The server integrates seamlessly with edge and cloud platforms, enabling real-time decision-making for AI applications. Solution simplifies NVIDIA NIM packaging and deployment with easy scaling without a need of advanced knowledge and experience.

​​Speakers:

​Jordan Nanos is a Machine Learning Architect who works as a Master Technologist in HPE’s North America CTO Office. He spends his time helping customers design and implement data-driven applications at any scale: from edge to cloud, sensor to tensor.  Day to day this means developing architectures to process data, develop models, and deploy them in diverse environments. Doing so requires extensive knowledge of the hardware, software, and services in HPE’s portfolio, as well as partners and open-source tools available in the ecosystem. Prior to this role, Jordan worked in HPE’s High Performance Computing organization as a Solution Architect, gaining hands-on experience measuring and tuning HPC system performance in the world’s most demanding datacenter environments.

Volodymyr Saviak is a Artificial Intelligence at Scale and High Performance Computing Sales Manager in Central Eastern Europe at HPE. For more than 25 years, delivers IT infrastructure solutions on the Central Eastern European market, focusing on High Performance Computing, Big Data, Storage and recently Artificial Intelligence at Scale. Volodymyr is helping organizations understand how to build and implement cutting-edge technical computing solutions focused on highest performance and scalability. For the development of the “First Ukrainian Supercomputer Center” and the Ukrainian HPC community, he was awarded Ukraine’s Parliament Highest Honor Award.