Greg Turner
Thirty years shipping software for Australian healthcare, government, banking and industry. Now designing and delivering the AI layer on top of it, from retrieval systems and agent workflows to the integration work that makes them usable inside a real organisation.
What I Do
Most AI work stalls somewhere between the demo and the day job. The model works, then nobody can get it past privacy, integration, cost or the people who have to use it. That gap is the work.
Retrieval augmented generation over your own content, LLM features inside existing products, agent workflows, and the evaluation and guardrails that make them safe to turn on.
Where the AI has to meet systems that already exist. Integration across REST, SOAP, message queues and legacy platforms, identity and access, cloud design that survives an audit.
You get me, not a team of juniors with an account manager in front of them. Senior-only delivery, accelerated by agentic tooling, inside your review and quality standards.
Selected Work
Seven engagements from the last decade. Each one shipped into production and, in most cases, is still running.
A first-to-market AI riding coach, built on retrieval augmented generation over the client's own rider training methodology. The interesting problem was not the chatbot. It was grounding: a coaching assistant that invents technique is worse than no assistant at all, so the answers had to come from the client's system and be traceable to it.
Delivered end to end in under two months, including the cross-platform mobile app and in-app payments for subscriptions and consumables on iOS and Android.
A complete Windows desktop application for managing costs on construction projects, designed and built solo from architecture through UI to release in C#, .NET and WPF with DevExpress components. Delivered one month ahead of schedule, using agentic development tooling without relaxing code quality.
SAML-based federated authentication for a C#/.NET application, establishing identity federation across university systems. SAML fails in ways that are difficult to observe, so the work included a mock identity provider to debug traffic before anything reached production, plus dependency conflict resolution through Fusion Logs and assembly binding configuration.
Shipped with a fully automated CI/CD pipeline on AWS CodeBuild and CodeDeploy, and launched into UQ's production environment.
Electronic delivery of internal referrals across Queensland Health. The constraint was that clinical criteria change constantly and a release cycle cannot keep up, so the forms are metadata-driven: criteria live as XML and JSON in SQL Server and render dynamic clinician-facing forms at runtime.
Modern Angular and C# services integrate with legacy clinical systems across both REST and SOAP boundaries, with RabbitMQ handling secure, scalable referral submission. Deployed across Queensland Health and still supporting clinical operations.
A Unity application for the Meta Quest headset allowing surgeons to control a surgical robot through an immersive VR interface, with real-time device interaction. Alongside it, a React Native app for surgical appointment management and a GraphQL backend serving both web and mobile.
Three years on healthcare interoperability: HL7-compliant FHIR APIs adopted across multiple client teams, and integration of Medicare Web Services and My Health Record into the Gentu patient management system. Claims processing cannot stop while you modernise it, and the compliance surface is not negotiable, so the work was as much about regulatory precision as engineering.
Cloud-native services on AWS Lambda and Docker across Fargate and EKS, with infrastructure as code in CloudFormation, CDK and Terraform.
A predictive reliability platform analysing factory plant models and work order histories to generate maintenance plans before equipment fails. Work order text is written by humans under time pressure, so the pipeline had to handle misspellings and inconsistent plural forms before any analysis was possible: AWS Comprehend for NLP, with a Python service for autocorrection and singularisation.
Serverless throughout, on AWS Lambda with MongoDB Atlas, delivered to industrial clients as a production platform.
"Greg is an absolute stand out. He's fast, knowledgeable and very easy to deal with."
About
I started as an analyst programmer in 1993 and never stopped writing code. Along the way I picked up the architecture side, a TOGAF practitioner qualification, an MBA and an AWS Solutions Architect certification, but the reason clients keep calling is that I still build the thing.
That combination is the point. Most AI engagements split into a consultancy that produces a recommendation and a delivery team that has to make it real, and the translation loss between them is where budgets go. I do both, which means the architecture is written by someone who knows what it costs to implement.
The work has mostly been in places where getting it wrong matters: Queensland Health clinical systems, Medicare claims processing, banking, the ATO, university identity infrastructure. Regulated environments teach you that a system is not finished when it works, it is finished when it survives an audit, an outage and the people who have to use it every day.
I'm based in Brisbane and work with clients across Australia and internationally.
Working Together
Option One
You have a team and a roadmap, and you need a senior engineer who can architect and build without a ramp-up period. Typically three to six months, day rate, on site in Brisbane or remote anywhere in Australia.
Option Two
You have a defined thing you need built: an AI feature inside an existing product, a mobile app, an integration between two systems that refuse to talk. Fixed scope, fixed price, delivered end to end.
Contact
A short description of the problem is enough to start. If it isn't something I'm the right person for, I'll say so and point you somewhere better.
