// Technology choices

Two layers.
One investment that lasts.

Every tool below is open-source or CNCF-aligned — chosen to protect your portability, your negotiating leverage with cloud providers, and the long-term ROI on the platform you're funding. No proprietary lock-in. No surprise relicensing.

AI Infrastructure Layer NEW
KServe / Triton
LiteLLM (Model Gateway)
LangChain / LlamaIndex
Weaviate / Qdrant / pgvector
MLflow / DVC
Ray (distributed AI)
Langfuse / Phoenix (observability)
MCP Server Framework
NVIDIA Device Plugin
Karpenter (GPU scheduling)
OpenCost (AI FinOps)
Ollama (self-hosted LLMs)
vLLM
Platform Engineering Layer (CNCF)
Kubernetes
Backstage
ArgoCD / Flux
Crossplane
Terraform / OpenTofu
Helm
Vault (HashiCorp)
OPA / Kyverno
Cilium / Calico
Prometheus + Grafana
OpenTelemetry
Tekton / GitHub Actions
cert-manager
Karpenter
Cloud & Infrastructure
AWS (EKS, Bedrock, SageMaker)
GCP (GKE, Vertex AI)
Azure (AKS, Azure OpenAI)
On-prem / Bare metal
Hybrid cloud
// Philosophy

Tooling is selected on merit, not on partnership economics — we hold zero vendor relationships that could shape our recommendations. Where managed AI services (Bedrock, Vertex, Azure OpenAI) are the right commercial fit we'll say so, while keeping a self-hosted alternative on the table for data-sovereign or regulated workloads. The result: a platform you can defend in a procurement review, an audit, and a budget cycle.

// Questions about the stack?

Need a defensible
technology choice for the board?

Not every tool fits every organisation. Let's talk about your context — and what you can stand behind in a procurement or compliance review.

Book an executive briefing →