Services

Kubernetes & container engineering.

We containerise apps and run Kubernetes where it earns its keep — EKS, AKS or GKE — with autoscaling and observability. NZ-based, working AU, UK and US.

Containers make software portable; Kubernetes makes a fleet of it manageable — but only when it is set up with restraint. We package your services in Docker, run them on the right managed cluster, and wire up autoscaling, health checks and observability so the platform stays quiet. We also tell you when Kubernetes is overkill and a simpler runtime would serve you better.

Where we help

Containerising existing apps with lean, reproducible Docker images and multi-stage builds
Standing up managed Kubernetes — EKS, AKS or GKE — with sensible defaults and guardrails
Helm charts and Kustomize so deployments are versioned, repeatable and reviewable
Horizontal and cluster autoscaling tuned to real traffic, not guesswork
Ingress, TLS, secrets and network policy set up so the cluster is private by default
Observability — metrics, logs and traces — wired in from the first deploy, not bolted on later
Resource requests, limits and right-sizing so you are not paying for idle nodes
Migrating from hand-managed VMs or bare Docker Compose to an orchestrated platform

Example builds

Data backbone services. vesseltrack.net ingests live AIS data and feeds FuelClock — the kind of always-on, multi-service workload that benefits from containerised deployment and clean scaling boundaries.

Full-stack platforms. On full-stack & data systems work we package web, API and worker tiers as separate containers so each scales and ships on its own schedule.

GPU and model workloads. Training and serving custom neural networks means containerising CUDA-heavy jobs and scheduling them onto the right nodes without starving everything else.

How we engage

We start by asking whether you actually need Kubernetes — plenty of workloads are better off on a simpler PaaS or serverless runtime. Where it fits, we use managed clusters, keep the YAML small and reviewable, and hand over something your team can operate. Security defaults come from our cybersecurity practice. See our services for how we scope and price.

Common questions

Do we really need Kubernetes?: Often not. If you run a handful of services with predictable load, a managed PaaS or serverless is cheaper and quieter. We recommend Kubernetes when you have many services, bursty scaling or genuine portability needs — and we will say so plainly either way.
Which cloud do you run Kubernetes on?: Whichever you are on — EKS on AWS, AKS on Azure or GKE on GCP. We lean on the managed control plane so your team is not patching masters, and keep workloads portable enough to move if commercials change.
Can you containerise an app we already have?: Yes. We write lean multi-stage Dockerfiles, pin dependencies for reproducible builds, and migrate you off hand-managed servers or ad-hoc Compose setups onto something orchestrated and observable.

Cloud & infrastructure CI/CD & GitOps Infrastructure as code

Considering this for a project?

Start a conversation

Where we help

Containerising existing apps with lean, reproducible Docker images and multi-stage builds
Standing up managed Kubernetes — EKS, AKS or GKE — with sensible defaults and guardrails
Helm charts and Kustomize so deployments are versioned, repeatable and reviewable
Horizontal and cluster autoscaling tuned to real traffic, not guesswork
Ingress, TLS, secrets and network policy set up so the cluster is private by default
Observability — metrics, logs and traces — wired in from the first deploy, not bolted on later
Resource requests, limits and right-sizing so you are not paying for idle nodes
Migrating from hand-managed VMs or bare Docker Compose to an orchestrated platform

Example builds

Full-stack platforms. On full-stack & data systems work we package web, API and worker tiers as separate containers so each scales and ships on its own schedule.

GPU and model workloads. Training and serving custom neural networks means containerising CUDA-heavy jobs and scheduling them onto the right nodes without starving everything else.

How we engage

Common questions

Do we really need Kubernetes?

Often not. If you run a handful of services with predictable load, a managed PaaS or serverless is cheaper and quieter. We recommend Kubernetes when you have many services, bursty scaling or genuine portability needs — and we will say so plainly either way.

Which cloud do you run Kubernetes on?

Whichever you are on — EKS on AWS, AKS on Azure or GKE on GCP. We lean on the managed control plane so your team is not patching masters, and keep workloads portable enough to move if commercials change.

Can you containerise an app we already have?

Yes. We write lean multi-stage Dockerfiles, pin dependencies for reproducible builds, and migrate you off hand-managed servers or ad-hoc Compose setups onto something orchestrated and observable.