Platform Engineer · Cloud, DevOps & Data
Systems Cartographer — I work at the intersection of software, cloud infrastructure, and applied ML, tracking what people actually want to build, then mapping the route from idea to running system: design, iterate, measure, then build, step by step, end to end.
I also run a small EU software venture (Estonia OÜ), where I handle product, infrastructure, and delivery solo. My current focus is raz, a blazing-fast Azure CLI written in Rust.
Push through the pain. Giving up hurts more.
Track what needs to be built and chart the route—application, infrastructure, architecture, and the boundaries between them—before writing the code that has to live with those decisions.
Ship maintainable software, APIs, CLI tooling, and cloud infrastructure. Rust, Python, Terraform, and Azure—picked for the job, not the résumé.
Design, iterate, and validate against real numbers—benchmarks, cost, reliability—so "it works" means end to end, not just on the demo.
raz—a blazing-fast Azure CLI, written in Rust—reimplements a slice of the Azure CLI (az) as a single native binary.
az is a Python application that re-initializes its interpreter and imports a large module tree on every invocation. raz starts in milliseconds and makes the same ARM/Entra calls with a fraction of the overhead.
What makes it different:
- Single native binary—no runtime to install, drops straight into a scratch container
- ~40× faster than
azon engine-bound commands (--version,--help, cachedaccount list) - Two front-ends over one core library: a
clapCLI and an interactiveratatuiTUI dashboard - Live ARM REST calls with long-running-operation polling; Entra OAuth device-code auth
- Honest benchmarks with the caveats spelled out—no cherry-picked wins
- azure-landing-zones-terraform — deploy, update, and destroy Azure Landing Zones with Terraform: staged plan/approve/apply, AVM pattern modules, drift detection, and privilege downgrade.
- iac-tf-azure — a reusable Azure Terraform module library.
- Caching per-tenant tokens in
razto close the live-ARM latency gap - Building an EU-registered software venture end to end—product, infrastructure, and ops
- Exploring applied ML and LLM tooling for real product workflows
Build things that are useful, understandable, and worth maintaining.


