Cloud, your data centre or both
We design environments in Azure, AWS or GCP, or on your premises. We compare total costs, including migration and administration, data location requirements and dependencies on systems you already use.
Infrastructure · Cloud · On-premises
We build the environment your applications run in.
Azure, AWS, your own data centre or a combination. We design infrastructure, set up deployment and monitoring, and prepare your team to manage it. We move systems into the cloud and back out.
What we help with
An environment for your applications and data
We start with what needs to run, the load you expect and how much downtime you can accept. This guides service selection, capacity planning and cost estimates.
Kubernetes and containers
We build Kubernetes environments in the cloud and in your data centre, and also work with OpenShift. We configure networking, storage, access and application deployment, including the update process.
Infrastructure as code
We define environments in Terraform, Pulumi or platform-specific tools. We keep changes in Git and apply them through pipelines so they can be reviewed and repeated.
Migration into and out of the cloud
We map dependencies, prepare the target environment and test data transfers. Before switching over, we agree downtime, checks and a rollback procedure in case problems arise.
Monitoring and recovery
We configure metrics, logs and alerts for the operations team. We set up backups and verify recovery. We document procedures for common operational problems.
Access, rules and costs
We establish permissions, change approval and ownership of individual services. We allocate costs by team or application so you can see what you are paying for and where usage is growing. We identify unused capacity and propose changes that account for availability and performance.
Infrastructure for AI
Where your AI will run
We run AI on our own GPUs and work with cloud models too. We select the environment for your solution based on its tasks, request volume, data handling and costs. The design also covers who will manage the models and infrastructure.
What we build with AI on this infrastructureModels in your own environment
We prepare GPU servers and model deployment based on capacity needs and licence terms. We check response times and behaviour under concurrent requests.
Access to models
Through an AI gateway, we configure application access to selected models, limits and usage records. We connect the rules to company identities and permissions.
Usage and operations
We track GPU utilisation, response times and model call costs. We adjust capacity and operational limits based on actual load.
How migration works
We prepare the move in advance
We adapt the process to your applications and operational constraints. Before the move, we agree who decides on the cutover and how we will confirm that it has succeeded.
Map the environment
We examine applications, data, networks and connections to other systems. We identify dependencies and constraints, availability requirements and current operating costs.
Prepare and test the move
We build the target environment and verify data transfers, access and application behaviour. A trial migration establishes the time needed and the issues still to resolve.
Switch over
We follow the agreed procedure and check data and application features. We prepare rollback criteria and steps in case the checks fail.
Verify operations and hand over
We check performance, logs, backups and costs. Your team receives documentation and practical training. We agree ongoing support and responsibilities together.
Our work
Infrastructure in practice
Cloud architecture and Kubernetes
Changes to core system architecture, DevOps practices and Kubernetes deployment. Infrastructure automation and service selection with attention to costs.
What we did for O2
Running Geocentrum in Azure
The address service runs in AKS. GitLab and Argo CD support deployment; Grafana, Loki and Azure Monitor provide monitoring.
What we did for Packeta Bizerba Váhy a Systémy s.r.o.Supporting AI in retail scales
Product recognition runs on the device. Azure supports image data, model training, version management and monitoring.
What we did for BizerbaWhat is holding your environment back?
Planning a migration, dealing with operational problems or preparing infrastructure for AI? We will review your current setup and propose next steps.
Contact us
Where could IT save you the most?
Tell us what costs you the most time, money or operational effort. Together, we will choose a first step and a way to measure its benefits.
info@cloudfield.cz