Principal Engineer at AWS in Melbourne, Australia 🇦🇺. I write up what actually works: fine-tuning local models, building simple things on AWS that scale well, and mucking around with my homelab.
Start here
Local models in mid-2026: the engineering that closed the gap
How open weights got close enough to run at home in 2026, the engineering that got them there, and why a memory crunch made the box pricier. read more →
Herding parallel agents on a remote box with herdr
herdr is an agent-aware tmux replacement that shows which parallel Pi or Claude Code session needs you. My remote-box setup and install notes. read more →
Building lgtmaybe: a PR reviewer for any model
How I built lgtmaybe, a provider-agnostic PR reviewer with keyless cloud auth, and the fiddly security and concurrency problems that were fun to solve. read more →
Latest writing
Designing MCP tools for agents
MCP tools advertise contracts to the model. Keep related filters typed and shape safe data in code; load large catalogs progressively. read more →
Why local models get stuck repeating JSON
A faster GPU turned 'too slow' into 'broken': why grammar-constrained JSON extraction sends local vision models into repetition loops, and what helps. read more →
Pydantic AI structured outputs and evals on Bedrock
How I combine Bedrock structured outputs, Pydantic Evals and a calibrated LLM judge to test model output in CI. read more →
Building and securing MCP servers with FastMCP
Build a FastMCP 3 server for company data. Add group-based access and an audit trail, then test the complete example offline. read more →
Skills or MCP servers: when you need a server
Skills can drive any CLI, but your laptop's credentials aren't an access model. Once data is shared and access has to be governed, you need an MCP server. read more →