Skip to content
Jack Burgess.

Search

Jump to a section or project, or run a quick action.

All projects

Developer platform

Streambased demo platform

A self-serve way to try Streambased: one install sets up Kafka, Iceberg and the product on Docker or Kubernetes, then a web app queries real-time and historical data together.

apps in one monorepo: web, API, agent, CLI and desktop
5
command to install the whole platform
1
desktop platforms: macOS, Windows and Linux
3

The challenge

To try Streambased, a prospect needed Kafka, Iceberg, object storage and the Streambased services all running and wired together — a lot to ask before a first demo.

What I did

  1. 1Built a desktop setup wizard in Electron and a terminal UI in Go (Bubble Tea) that check Docker or Kubernetes is ready, show exactly what will be created, then deploy the platform into its own isolated namespace.
  2. 2Wrote a Go agent that provisions and manages the demo environment — Kafka, Iceberg on MinIO, the Streambased services and generated logistics data — through the Docker and Kubernetes APIs.
  3. 3Built the web app in TanStack Start on a Hono API: a SQL editor that queries Kafka and Iceberg together, notebooks, charts, and browsers for Kafka topics, schemas, consumer groups and storage buckets.
  4. 4Added an AI assistant that answers product questions and points people to the right page, plus an MCP server so AI agents can read schemas and run queries.
  5. 5Set up a tag-triggered release pipeline that builds multi-architecture Docker images, cross-compiles the CLI, packages the desktop app for macOS, Windows and Linux, and publishes checksums the installer verifies.

The outcome

Anyone can now go from one install command or download to querying Kafka and Iceberg together in the browser, on their own laptop or cluster, without running a docker or kubectl command by hand.