You can watch it run
Observability and evaluation go in from the start, not just before handover. You never have to take our word for whether something works.
The practice
A small, senior engineering practice. We design and build cloud platforms, and the AI systems that run on top of them, for teams who need both to keep working after we leave.
Where the work comes from
It started on datacenter floors. Each shift since has been built on top of the one before it, not in place of it.
- First
Datacenters
UNIX, Linux and FreeBSD servers, the databases and networks between them, and the pipelines that moved data across all of it.
- Then
Cloud
The same estates moved onto managed platforms: infrastructure as code, Kubernetes, delivery paths, and the networking and identity that hold them together. Mostly Google Cloud, with estates on AWS and Azure as well.
- Now
AI and agents
Agent systems and the platforms they run on, with evaluation and observability built in from the start. They only work because the layers underneath them do.
Fig. 9Why the order matters
Nothing above replaced what came before it. The agent systems we build now sit on the same platform work, which sits on the same estate work, and the dashed lines are the part most teams skip: a layer that is not registered to the one under it is a layer that moves independently, which is another way of saying it falls.
- 108Foundation
- 220Framework
- 336Modules
- 452Registration
Two things we do
Every project we take on comes down to one of these two, or to both.
We integrate.
We see how systems that were never designed together connect: a cluster, a warehouse, a queue, a vendor API, the spreadsheet somebody emails every morning. Then we write the code that makes them behave as one.
Systems that were never designed together.
We systemise.
We take a process that works but will not scale, find the structure inside it, write that structure down, and hand it to machines.
A process that works and will not scale.
Our principles
Three principles. We’d rather keep three than list ten.
Built to be handed over
When the structure is right, a system is easier to run, change and explain. We build for the team who inherits it, which is usually yours.
Nothing spare
Every part should be earning its place. If it isn’t doing real work, we take it out rather than leave you maintaining it.
How the practice runs
The short version, for anyone deciding whether to call.
- Size
- Small. One lead, with collaborators brought in project by project.
- How work arrives
- By referral, not advertising.
- What we take on
- Whole projects with a clear outcome and an owner on your side.
- Where we work
- Inside your own repositories and cloud accounts, and on site when the project needs it.What forward-deployed means here
- Platforms
- Google Cloud first. AWS and Azure estates as well.
- Google Cloud
- Member of the Google Cloud Partner Network.
- While it runs
- A written update every Friday: what moved, what didn’t, and what we got wrong.
- At the end
- A handover your team can run without us.
- Clients
- Never named. Every project on this site is anonymised.
- Certifications
- Google Cloud Professional Cloud ArchitectGoogle Cloud Generative AI LeaderCNCF Certified Kubernetes Administrator
- Registered
- A California limited liability company.
Where we fit
The same three statements, said both ways. Read either column; they are the one filter, and it is said up front so neither of us spends a call finding out.
Where we do our best work
What we turn down
01Whole projects with a clear outcome and an owner on your side.
01Staff augmentation. We take on projects, not seats on your team.
02Problems where the blocker is technical and the decision has already been made.
02Work where the thing that won’t scale is a person, or a decision nobody has made yet.
03Teams who want to run and extend what we build after we hand it over.
03Anything where we would be the only people who understand it at the end.
If that sounds like the practice you’re looking for, the next step is one call, about ninety minutes.
Saguaro Cloud Systems is led by JR Camou, with collaborators on each project.