Deploying AI infrastructure shouldn't require a PhD and a three-week onboarding. Yet here we are, in 2026, where spinning up a local LLM, connecting it to MCP tools, and managing the whole thing with a sane interface still feels like assembling IKEA furniture with a crayon.
We built ChurroStack to fix that.
The problem we were solving
At IBERANT, we run a lot of internal infrastructure. Local LLMs for testing. MCP (Model Context Protocol) servers for our agents. Observability tools. Knowledge bases. The usual chaos of a software company that builds AI products.
Every time we onboarded a new engineer, the setup process consumed most of a day. Every time we spun up a new project environment, we were copying YAML files around like it was 2017. And every time someone asked "how do I add a new MCP tool?", the answer involved four config files, a Docker network, and an unspoken prayer.
"Infrastructure should be as simple as the product it enables. We kept saying that out loud, until we finally decided to build the thing that made it true."
What ChurroStack does
ChurroStack is a self-hosted platform for deploying and managing:
- Local LLMs — Ollama, vLLM, and custom model backends
- MCP servers — the emerging standard for connecting AI agents to tools and data
- Application services — databases, caches, API gateways, message queues
- Developer tooling — observability dashboards, log aggregation, health monitoring
You get a complete AI development environment running locally or on your private server — LLMs, tools, and all — in under five minutes.
Under the hood, ChurroStack uses Kubernetes for orchestration, Istio for service mesh, and ssh for reverse tunneling. Nothing exotic. Everything predictable. Everything debuggable when something goes wrong.
Why Open Source?
We believe the infrastructure layer of AI development should be a commodity. The value isn't in managing YAML files — it's in what you build on top of them.
By open-sourcing ChurroStack, we accomplish three things:
- We give back to the community that made our own products possible. Every library, every tool, every open standard that we've built on deserves reciprocity.
- We build trust through transparency — you can read every line of code that runs your infrastructure. No black boxes, no unexplained behavior.
- We improve faster — contributions from the community expose use cases we never considered and fix bugs we never found. This makes ChurroStack better for everyone.
Getting started today
ChurroStack is available now at churrostack.com. The project is open source and we welcome contributions, bug reports, and feature requests.
If you're building AI systems and tired of managing infrastructure by hand, ChurroStack is for you. Give it a try, star the repository, and tell us what you think.
The infrastructure problem has been solved. Now you can get back to building the thing that actually matters.