Flowmail
An AI email productivity platform on Cloud Run with Supabase, integrating Gmail via OAuth. A Gemini-powered feature suggests board organization and automated rules with a human-in-the-loop accept/reject workflow.
The full catalog, not just the highlights: AI tooling, production apps, systems software, and the tooling that keeps it all honest.
An AI email productivity platform on Cloud Run with Supabase, integrating Gmail via OAuth. A Gemini-powered feature suggests board organization and automated rules with a human-in-the-loop accept/reject workflow.
A working prototype of an LLM tool-calling agent that answers natural-language questions about a Drupal site's content and configuration, backed by a crawler that builds a searchable SQLite knowledge graph.
A live property management app in daily production, designed and shipped solo. A double-entry ledger tracks balances, with tenant KYC verification and row-level-security private storage.
A native macOS menu bar app in Swift that exposes running apps and system state over a local WebSocket server, paired with a React Native companion app for real-time switching over the local network.
A local-first desktop app for developers with hundreds of repos on disk: it scans your machine, gives every project a landing page with git status and README, tracks how much space node_modules and build/target folders are wasting, and cleans them safely through the OS trash. A cross-platform Tauri rewrite of a native macOS app I built first in Swift.
A fully on-device personal finance app for Android that reconciles two capture paths, real-time SMS parsing and a one-tap quick-entry prompt, against a monthly bank PDF or CSV statement parsed natively on the phone. Every parser, the database, and the reconciliation engine run locally; no server, no cloud sync, financial data never leaves the device.
A self-hosted visual regression testing tool built on top of BackstopJS, with a GitHub Actions pipeline that crawls a sitemap and deploys reference/test screenshots to Vercel. Built to catch visual regressions before they ship. This kind of automated gate is what let my teams push deployment velocity up 40% without trading away release confidence.