AI Solutions Camp stacks

Self-hosted AI stacks you can run on your own infrastructure. Every stack ships the same set of deployment options and is tested in CI:

Stack	What it runs	Compose	Helm	Terraform (Docker)	Terraform (AWS)
RAG Pipeline	Retrieval-augmented generation over your documents	✓	✓	✓	✓
LLM Chatbot	Private chat UI with local models	✓	✓	✓	✓
Embedding Service	Text embeddings over HTTP	✓	✓	✓	✓
AI Agent Framework	Visual builder for LLM agents	✓	✓	✓	✓

Each stack is also listed with a guided setup on aisolutionscamp.io/stacks.

Layout

stacks/<slug>/
  README.md               documentation for the stack
  stack.json              catalog metadata (name, description, tags, env variables, ...)
  compose.yaml            Docker Compose deployment
  .env.example            settings, copy to .env
  smoke-test.sh           starts nothing, only checks a running stack (used by CI)
  helm/install.sh         Helm commands
  helm/values.yaml        values for the chart
  terraform/local/main.tf Terraform with the Docker provider (single machine)
  terraform/aws/main.tf   Terraform for AWS
  quickstart/*.md         short setup notes per deployment option

What CI checks

• docker compose config for every stack, then docker compose up and smoke-test.sh against the running containers
• helm lint / helm template (or the chart install command for upstream charts)
• terraform fmt -check and terraform validate for both Terraform variants

The AWS Terraform is validated but not applied in CI, because applying it creates billable resources. Review it and run terraform plan in your own account first.

License

MIT