Data & application platforms
Composable, cloud-native architecture on Java, Scala, Spark, Delta Lake, Azure and Kubernetes — designed as platforms, not one-off systems.
James Baxter — Software Architect & Principal Engineer · Brisbane, Australia
/ Intro
For over 20 years I've built software that grew large enough, or complicated enough, that the obvious answers stopped working. My background spans hands-on engineering, architecture and engineering leadership in enterprise product companies, with real depth in distributed JVM systems, cloud-native platforms and data-intensive applications.
Today I work with a global industrial-equipment manufacturer, where I proposed and led the replacement of the core analytics pipeline behind a large-scale industrial IoT platform — work that has since grown into a reusable data platform supporting multiple product lines. I'm drawn to projects where the boundaries are unclear and the consequences matter.
/ What I do
Composable, cloud-native architecture on Java, Scala, Spark, Delta Lake, Azure and Kubernetes — designed as platforms, not one-off systems.
Distributed systems that have reached their architectural limits, re-shaped so they can grow again without a rewrite-the-world bet.
Failures that resist conventional debugging — traced all the way down to the JVM, the dependency graph or the packet when they need to be.
A full stack that runs from a single command, and AI-assisted, specification-driven workflows that accelerate delivery without giving up architectural control.
/ Selected work
Proposed and led a data-platform rebuild that survived an organisational pause and came back as the foundation for multiple product lines.
Doubled the throughput of an enterprise product by finding and fixing the filesystem and database bottlenecks that were really constraining it.
Grew an engineering function from a single developer into a ten-person cross-functional team — and the delivery practices to match.
Traced a run of intermittent gateway 504s that never reached the application logs down to a single bit-flipping TCP timestamp option corrupting packet checksums — found with packet capture and Wireshark.
Built a specification-driven SaaS platform with AI-assisted delivery, cutting implementation time dramatically while keeping full architectural control.
Published VendKit, an open-source Go tool that vendors curated slices of files across repositories — kept faithful to upstream by a checksum manifest, upgraded through review-gated pull requests. github.com/jameswbaxter/vendkit ↗
/ About
Away from the keyboard: motorbikes, and reading about behavioural and macroeconomics.
/ Contact
The fastest route is email — happy to talk through a platform, a modernisation, or a failure nobody can explain.
For consulting engagements — AI delivery systems, platform architecture and modernisation, production investigation — see baxter.consulting ↗.