Data and platform engineering at Omega Point. Austin, TX.
I spend my days in the unglamorous layer between a data platform and the people using it — Databricks jobs, Airflow DAGs, risk-model pipelines — and most of what I publish is tooling to make that layer legible from a terminal.
Nearly everything here follows the same shape: one CLI an agent can drive, and one TUI a human can read. The CLI is scriptable and composable so a coding agent can call it without a browser; the TUI is for the five minutes when you actually want to look at something.
- databricks-tools —
dbtoolsfor jobs, runs, notebooks and Unity Catalog;dbtuito browse them read-only - dev-tools — live terminal dashboards for GitHub PRs, Airflow, and Azure DevOps pipelines
- goblin-watcher — parallel coding agents in git worktrees: ticket → branch → worktree → session
- hc-cli —
hcfor Healthchecks.io: the full v3 management API, plus outgoing pings for whatever cron job you're babysitting - spg — per-project commands published to
~/bin, with zsh completion from the same declarations
Python on uv, just
for tasks, ty and
ruff keeping me honest, asdf pinning the
bootstrap CLIs. Heavy Claude Code user — the tools above mostly exist because
agents need a good CLI even more than humans do.