Services and fixed-price packages

An Azure or AWS foundation written in Terraform: networking, identity, logging and cost controls.

A cloud account you can grow into, set up in your first month instead of your first crisis.

A typical project

A seed-stage team has been deploying to an AWS or Azure account by hand, with shared logins and no record of what is running. We rebuild the foundation as code, separate development from production and add cost alerts, so your first hires inherit something safe.

A cloud foundation built around who can see what, with the evidence to show it.

A typical project

A small firm is moving a line-of-business system to the cloud and needs to explain to clients where data lives and who can reach it. We design the network and access model first, log everything that matters and document the data flows for your data protection lead.

A repeatable cloud baseline you can offer to more than one client.

A typical project

An IT provider keeps rebuilding similar Azure or AWS setups for small clients, each slightly different. We turn your best version into a parameterised Terraform baseline, so the next client takes days instead of weeks.

A hardened cluster with ingress, TLS and monitoring, ready for your first workloads.

A cluster your team can deploy to on day one, without hiring a platform engineer.

A typical project

A product team has outgrown a couple of servers and deploys containers by hand. We stand up a managed cluster, set up ingress and certificates, and give developers one clear way to ship a service.

Containers for reproducible research, on the hardware you already have.

A typical project

A group wants to run analysis pipelines and models in containers so results can be repeated, on a small on-premises cluster or a few GPU machines. We build a cluster that fits the hardware and the way researchers work.

A hardened cluster you can run for a client without owning every decision.

A typical project

A provider has a client who wants Kubernetes, but nobody on the team has run it in production. We build and harden the cluster to a written baseline and leave runbooks your engineers can follow.

Automated build, test and deploy, so releases stop depending on one person's laptop.

Releases that take minutes and do not depend on one person.

A typical project

A small team deploys by copying files or running scripts from a laptop, and every release is a stressful event. We build a pipeline that tests, packages and deploys on every merge, with a clear way back if something breaks.

A pipeline template you can reuse across client projects.

A typical project

An IT provider supports software for several clients, each deployed differently. We build one standard pipeline with client-specific settings, so builds, tests and deployments follow a single process.

Ansible roles that build, patch and harden your Linux servers the same way every time.

One tested way to build and patch every research server, whatever its distribution.

A typical project

A research computing team looks after a mix of Rocky, Red Hat and Ubuntu machines built over several years, with scientific software installed by hand. We capture the standard build in Ansible so new nodes match, and patching becomes planned instead of improvised.

Patching and hardening you can show an auditor.

A typical project

A firm has to evidence that its servers are patched and configured to a standard, and currently relies on spreadsheets and memory. We write the standard into Ansible and produce the reports from the same run.

Fewer one-off fixes and a cleaner patching routine across client estates.

A typical project

A provider patches client Linux servers by hand, with different scripts for each client. We build shared Ansible roles with a separate inventory per client, so patching runs the same way everywhere and reports are produced automatically.

Slurm scheduling, shared storage and GPU support for a research group or lab.

A shared compute system that scientists can use and your team can run.

A typical project

A group has bought or been given GPU servers and a storage array, and everyone is logging in with shared accounts and competing for the same cards. We set up Slurm with fair-share rules, a storage layout with quotas and the software environments people need.

A language model running on your own infrastructure, tested on one real use case.

Try AI on client documents without those documents leaving your control.

A typical project

A practice wants to summarise and search long documents, such as contracts or case files, but cannot send them to a public AI service. We run a model inside your environment, limit who can use it, log what it does and test it on one real task.

Build an AI feature on infrastructure you control, before you commit to a vendor.

A typical project

A startup wants an AI assistant over its own data, but customers keep asking where their data goes. We deploy an open-weight model, test retrieval over your content and show what quality and cost look like with self-hosting.

Run language models on unpublished or restricted data, on hardware you own.

A typical project

A group wants to use a language model on unpublished results or restricted datasets that cannot leave the institution. We deploy a model on your GPU servers, connect it to a collection of papers or data descriptions and test it on one research task.