Four New Addons: PenTest, Staging Clusters, and GPU Infrastructure

We're expanding the Aureus addons marketplace with four new offerings — each one addressing something we've heard from customers repeatedly. Here's what's new and why we built each one.

Penetration Testing — $499 per engagement

Security testing is one of those things that most teams know they should do and consistently defer. The usual blockers: finding a vendor, scoping the engagement, waiting weeks for a report, then figuring out what to do with findings that weren't written for engineers who have to actually fix them.

The Aureus PenTest addon removes that friction. Because we already manage your cluster, we know your network topology, your application architecture, and your deployment configuration. We don't need an onboarding call to understand what we're testing. We can scope and execute a targeted engagement in days rather than weeks.

The engagement covers:

• External and internal network perimeter testing
• OWASP Top 10 application surface testing
• Authentication and session management review
• Privilege escalation and lateral movement checks
• A written report with CVSS severity ratings and remediation guidance specific to your stack
• A follow-up call to walk through findings

At $499 per engagement this is significantly below what you'd pay an external vendor for the same scope — and the output is actually actionable because it's written by the same team that will help you fix what it finds.

Staging Clusters — $399/mo (Cluster S) · $699/mo (Cluster M)

Deploying directly to production because you don't have a staging environment is one of those technical debt decisions that feels acceptable until it isn't. The first time a schema migration or a dependency update takes down production, the math changes quickly.

Staging clusters from Aureus are not a shared namespace or a lightweight simulation. They are dedicated bare metal nodes that mirror your production Kubernetes configuration with full network isolation. Separate hardware, separate database instances (with sanitised seed data by default), and separate deployment pipelines — managed by the Aureus team alongside your production environment.

The practical implication: you can run migration rehearsals against real data volumes, execute load tests without affecting production metrics, and deploy release candidates with confidence. Any change that would cause a problem in production will surface in staging first.

Pricing is tier-matched: $399/mo for Cluster S environments, $699/mo for Cluster M. The delta reflects the HA control plane replication on Cluster M, which is necessary to reproduce the behaviour of a multi-node production setup accurately.

GPU Cluster Setup — $2,000 one-time

More and more teams are coming to Aureus specifically for AI/ML infrastructure — model serving, fine-tuning runs, embedding pipelines, vector search at scale. The base Aureus offering handles the application and database layer well. GPU workloads require additional setup that deserves its own scope.

The GPU Cluster Setup addon covers the full provisioning of a GPU-backed node on your cluster:

• Bare metal GPU node provisioning and hardware verification
• CUDA or ROCm driver installation and validation
• NVIDIA device plugin integration with your Kubernetes cluster
• MIG (Multi-Instance GPU) partitioning for A100 and H100 hardware
• Deployment of your first GPU workload — whether that's a vLLM inference server, a Triton deployment, or a custom training job
• GPU monitoring via Grafana with DCGM exporter (GPU utilisation, memory, temperature, error counters)
• Runbook and handover documentation

There are a few things worth being specific about here. GPU Kubernetes setup is genuinely non-trivial. The NVIDIA device plugin configuration, getting the right CUDA version aligned with your model's requirements, configuring MIG correctly for your workload mix — these are the kinds of problems that take a competent engineer a day or two to get right the first time. The $2,000 one-time fee covers that work so you don't have to.

If you're running inference workloads, we'll configure the deployment for your serving framework of choice (vLLM, Triton, Ollama). If you're running training jobs, we'll set up the job scheduling configuration and resource limits. Either way, you get a cluster that's properly set up for GPU work, not just one that has a GPU attached to it.

Pricing and availability

All four addons are available immediately from the addons marketplace in the customer dashboard. The penetration test and GPU cluster setup are one-time engagements. Staging clusters are billed monthly and can be cancelled at any time.

If you have questions about any of these — scope, timing, or whether they're right for your specific setup — open a ticket from the dashboard or reach out directly. We're happy to talk through it before you commit.