Halyard
2025A queue-backed ingestion pipeline handling 40k events/sec with end-to-end tracing, backpressure, and zero-downtime deploys.
A selection of backend systems, developer tools, and the occasional full-stack project. Each one taught me something about reliability, observability, or simply writing code that's kind to the next person who reads it.
A queue-backed ingestion pipeline handling 40k events/sec with end-to-end tracing, backpressure, and zero-downtime deploys.
Double-entry accounting core with a typed API, deterministic tests, and a clean audit trail by design.
An open-source observability dashboard that turns noisy spans into a story engineers actually read.
A schema-migration tool with safe rollbacks and a dry-run planner that shows exactly what will change.
A feature-flag and config service with audit history, gradual rollouts, and sub-millisecond reads at the edge.
A self-hosted job runner with fair scheduling, retries with jitter, and a tiny, legible web UI.