Nebius AI monthly digest, November 2023

This is the first digest since access to our AI-centric cloud platform was unlocked just a month ago. Even during such a short time, we’ve published some exciting materials that we’re eager to share with you.

Platform and productsPlatform and products

• Nebius AI walkthrough
  Dive into the platform with a walkthrough video featuring Nikita, a Product Manager at Nebius AI, giving a closer look at the platform’s features and service management:

• Starting a Kubernetes cluster in 6 minutes
  Check out a quick video tutorial by Boris, our Solution Architects Team Lead. Setting up a Kubernetes cluster quickly is a process you might need to master for autoscaling your fleet of machines exactly when required.

• Kubernetes and its significance for ML/AI engineers
  Get even deeper into the K8s with Levon, who elaborates on the properties that have made Kubernetes a de facto industry standard for orchestrating. There are two main criteria and two more including a handy addition of Terraform — all of which are uncovered in a blog post.

• Running PyTorch on Nebius AI virtual machines
  Nikita is back, this time discussing PyTorch. Learn how to master the swift process of deploying machine learning workloads on our platform. Please feel free to comment on YouTube if you have any technical questions for Nikita.

Hardware and supercomputerHardware and supercomputer

• ISEG supercomputer: 16th in the world and 4th in Europe
  Discover more about ISEG. We weren’t aiming to create a supercomputer, yet our R&D team decided to test a part of the platform, which was free of customers’ workloads at that moment. For that, they used a benchmark from the Top 500.

• Our data center and hardware
  Explore our own DC in Finland, which reflects the company’s dedication to maximum power efficiency. Modern solutions of the whole data center allow us to continuously scale our GPU-equipped fleet with green agenda in mind.

• Designing servers and racks to host modern GPUs
  Our latest server rack generation presents two distinct node solutions: one custom-built for training and the other optimized for inference. But why are GPUs and AI so inextricably linked? Find out in our blog post.

• H100, A100, and others: choosing the right graphic card
  If you’re deciding which graphics card to choose, check out an article by Igor, our IaaS Technical Product Manager, for a quick answer. He breaks down the differences between NVIDIA’s GPUs and identifies workloads where each card performs at its best.

We plan to continue delivering insightful articles, how-to videos on using the platform and related tools, as well as other formats. To stay tuned, be sure to subscribe below.