Building AI systems in Elixir.
RSS FeedEngineering at the intersection of AI and distributed systems. Writing about LLM evals, agent architectures, and Elixir.
Featured
SKILL.md owns the flow, scripts own the work, state files own the trust, artifacts own the review, README owns the context
A design framework for agent skills that hold up under real use: five responsibilities, each assigned to exactly one place, none of them the agent's prose output.
Recent Posts
Building mix hex.audit_cvss: CVSS-Scored Dependency Auditing for Elixir
How I built a Mix task that pulls pre-computed CVSS 4.0 scores from Hex.pm's API to give Elixir projects a CI-friendly, multi-format dependency security audit — and what I learned about the gap in existing tools along the way.
System-Wide Logging for Security and Audit in Elixir/Phoenix
How we added structured access and action logs to a Phoenix/LiveView app — with trace ID correlation across HTTP and WebSocket, Okta identity propagation, and zero per-handler boilerplate.
Building This Blog
How I set up this blog — from cloning a template to publishing on GitHub Pages.