OUR PARTNERS:
Technical debt is the bane of most businesses and this is something that fit for purpose, design and Architecture principles can help mitigate. Whether it’s a public cloud solution or an on premise (private cloud) solution, we have you covered. As a team we have been designing and building platforms for AI, ML, & HPC for 5x years both in public cloud and private cloud with great success.
Our foundational understanding of Compute, Storage, Networking, Data, Security, Containers, Kubernetes and Observability, allow us to design applications and platforms from the ground up that are fit for purpose. We are well versed in CI/CD (Continuous Integration & Continuous Delivery), Kubeflow and MLOPs, so we understand pipelines and the required process and associated Architecture that is required to deliver the best possible results for the most effective price. We have the ability to provide full discovery workshops on a customers requirement and the Data landscape which gives us the valuable insight we need to design the most suitable platform.
In collaboration with MergeIT, CodeZero is the first partner in the world to design a specific AI/ML/HPC cluster on Dell servers, Intel CPUs and NVIDIA GPUs, which is fully funded by Intel. This cluster has been specifically designed for Immersion cooling and has a VMware software overlay that utilises Software Defined Storage (vSAN), Virtualisation (vSphere) and Kubernetes management (Tanzu) and cohesively works with NVIDIA’s AI for Enterprise software stack.
We have a number of validated reference architectures that we have run benchmarks on as part of ‘Project Demeter’ and we will be consistently expanding these solution stacks. Our design principles go deeper than just designing servers and Architecture, when it comes to cluster design, we can go as far as assigning a specific process to a specific node. This means instead of sharing resources across a server cluster, we are designing nodes by purpose and process. Microservices is widely seen as the next generation application Architecture, this allows a business to deploy multiple Containers and Kubernetes clusters which provides a Distributed Architecture meaning each cluster can form part of an application and can be scaled independently. For our AI/ML/HPC clusters we are taking a similar approach but it’s more like reversed, centralised Microservices as the cluster and processes are localised.
It’s these types of design principles along with our experience, knowledge, partnerships and proprietary software that separates us from the competition.
OUR PARTNERS: