Start Here

I am Uday Kapur, a Data Engineer at Aurecon in Melbourne, Australia. I build cloud data platforms, the foundations that make analytics, automation, and AI adoption possible. I am most useful when the problem is structural: messy ETL/ELT pipelines, missing data quality gates, or enterprise data platforms that weren't designed for what they're now expected to do.

What I work with

fabric · databricks · snowflake · dbt · spark · python · sql · azure · gcp

Most of my work is on cloud data platforms: designing them, rebuilding them, or making them reliable enough that downstream teams can stop worrying about whether the numbers are right. That covers data modelling, data governance, CI/CD for pipelines, and building analytical datasets that hold up under production load.

More recently, I have been building agentic development workflows that use AI agents constrained by project-level harnesses to produce artefacts at scale. Loop engineering, not prompt engineering.

Where to start reading

What Makes a Data Platform AI-Ready is the best overview of how I think about platform work. If you're more interested in the engineering detail, Why Your dbt Project Will Become Unmaintainable and Test Harnesses for Data Pipelines are closer to the ground.

For Microsoft Fabric and agentic workflows, Why Your Semantic Model Does Not Belong in a PBIX File covers the architecture of a 25-report Fabric project, and Loop Engineering for Microsoft Fabric covers the AI-assisted development workflow that built it.

For sector-specific work, the mining and energy articles cover real integration problems with Maximo, SAP PM, OSIsoft PI, and OPC. The real estate case study shows what happens when you apply loop engineering outside data platforms entirely. The Stripe MCP server shows the infrastructure side: building AI-safe developer tooling with PII redaction and input validation derived from source types.

Get in touch

If any of this matches what you need, the best routes are on the contact page.