ETHUX

Run AI models on your terms.

On your own hardware, or ours. We deploy the model, the cluster underneath and the tool your team uses. NVIDIA certified.

Services

Four things we build.

On-prem AI

LLMs running on your GPUs.

  • vLLM, SGLang and Ollama
  • RAG with local vector stores
  • Fine-tuning on your hardware
  • OpenAI-compatible routing
  • GPU monitoring and capacity planning

Custom LoRAs

A model tuned to your data, without training one from scratch.

  • Training data prepared from your documents
  • LoRA training on our GPUs or yours
  • Measured against the base model
  • Served on our API or your own hardware
  • The adapter is yours to keep

Private cloud and hybrid infra

The cluster underneath the AI, on your hardware or mixed with cloud.

  • Hybrid routing: sensitive requests stay local, the rest goes to the cloud through Erebus
  • Talos Linux and Kubernetes
  • FluxCD GitOps
  • Proxmox and GPU passthrough
  • Grafana, Prometheus, Loki
  • Self-hosted GitLab and CI runners

Custom apps

Software for what off-the-shelf cannot do.

  • FastAPI services and internal tools
  • Next.js dashboards and admin panels
  • n8n workflow automation
  • Self-hosted SaaS replacements
  • Integrations between your tools
The stack

The tools we use.

InferencevLLM, SGLang, Ollama, LMCache
ModelsQwen3.8-27B
OrchestrationTalos, Kubernetes, FluxCD, Helm
VirtualisationProxmox, GPU passthrough, cloud-init
AutomationOpenTofu, Ansible, n8n
CI/CDGitLab, GitLab CI
ObservabilityGrafana, Prometheus, Loki, ClickHouse
App layerFastAPI, Next.js, Postgres, Redis
PrivacyErebus, Headscale