Containers and orchestration
Helm
Package manager for Kubernetes applications. Helm is the package manager for Kubernetes, a graduated Cloud Native Computing Foundation project. A Helm chart is a set of templated Kubernetes manifests with configurable values, which can be versioned, shared and installed as a release. Most third-party Kubernetes software, from databases to monitoring stacks, is distributed as Helm charts.
When it is the right choice
- You deploy the same application to several environments with different settings
- You install third-party software such as ingress controllers or monitoring on Kubernetes
- You want versioned releases with upgrade and rollback
- Heavy templating can become hard to read; for simple overlays Kustomize may be clearer
How we use it
- Write clear charts for your services with sensible defaults and per-environment values
- Deploy charts through Argo CD so the cluster always matches what is in Git
- Lint and render charts in CI to catch mistakes before they reach a cluster
- Track upstream chart updates and upgrade on a schedule
Kubernetes
Argo CD
Prometheus
GitHub ActionsFrom the blog
CI/CD for Models and PromptsEvaluation gates, model registries, canary releases and rollback: applying CI/CD discipline to prompts and models in LLM applications.
GPU Nodes on Kubernetes Without Wasting MoneyNode pools, MIG, time-slicing, scale to zero and spot instances: practical ways to keep Kubernetes GPU costs under control for AI workloads.
vLLM vs Ollama vs TGI: Choosing an Inference ServerHow vLLM, Ollama and Hugging Face TGI differ in throughput, hardware support and operations, and which one fits your workload.
Related articles
CI/CD for Models and PromptsEvaluation gates, model registries, canary releases and rollback: applying CI/CD discipline to prompts and models in LLM applications.GPU Nodes on Kubernetes Without Wasting MoneyNode pools, MIG, time-slicing, scale to zero and spot instances: practical ways to keep Kubernetes GPU costs under control for AI workloads.vLLM vs Ollama vs TGI: Choosing an Inference ServerHow vLLM, Ollama and Hugging Face TGI differ in throughput, hardware support and operations, and which one fits your workload.Questions
Helm: frequently asked
Helm or Kustomize?
Helm suits reusable, parameterised packages and third-party software. Kustomize suits small patches on plain manifests. Many teams use both.
Should we run helm install by hand in production?
We recommend not. Deploying charts through GitOps with Argo CD gives you review, history and drift detection.
Running Helm, or deciding whether to? Tell us what you have; the first consultation is free.
Let's talk »
