Projects

An in-memory serving layer in front of BigQuery that gives dashboards sub-second latency while sharply cutting BigQuery slot cost — Looker Studio reads from it instead of hitting BigQuery directly.

A multi-tenant Big Data & AI platform (serving thousands of companies) that ingests, transforms, and serves data through analytics, Insights, and Smart Apps. I led its SQL-first engine migration — ~70% lower platform cost, 16× faster.

A 0→1, offline-first face-recognition time clock I took from concept to production in under four months — now serving hundreds of thousands of workers daily, protected by a granted U.S. patent.

A Change Data Capture (CDC) tool that ingests data from virtually any source — JDBC databases, MongoDB, and files — into the Data & AI Platform, online or in batch.

A batch pipeline that loads ERP data from a Google Cloud Storage folder into the Carol data platform — auto-discovering what to load and generating the BigQuery SQL from a single tenant id.

A scalable billing system that computes cost across cloud providers and splits it among tenants via custom rules — plus automated actions that actively reduce cloud spend.

Reverse ETL that lets SQL pipelines sync data into transactional databases — for applications that need a transactional store for scalability, performance, or transactions.

A cross-cutting program to cut data-platform cost across storage, ingestion, and processing — driven by architecture reviews and impact-vs-effort prioritization, reducing spend by millions without compromising performance.

My take on Microsoft Money, rebuilt for today: personal finance with honest multi-currency net worth from exchange rates synced daily, a modern interface, and AI that sets up accounts from plain words and reads bills from email.

Built to help my 12-year-old find sheet music he couldn't get anywhere: upload a recording, pick one of 22 instruments, and an AI pipeline writes the score at three difficulty levels — with synced playback and PDF / MusicXML / MIDI export.

A family app that turns everyday responsibilities into measurable independence — chores, allowance, habits and reflection, with an AI coach for parents and kids.

A home-maintenance app that turns everything a house needs into an auto-generated, remind-me-only-when-due plan — live on iOS and web, backed by an Express/Postgres API on Cloud Run.

A text optimizer that helps you catch spelling errors and redundancies before sending important messages — not a keyboard replacement, but a focused tool for clearer communication.

This site — a Next.js app on Firebase with a CMS-style admin, image uploads, and a Claude-powered writing assistant. A living portfolio I fully own and edit in place.

A paste-and-go web app: drop in SQL and instantly get a pretty-printed version, an interactive AST view, and a checklist of query-optimization warnings across a dozen dialects.

Built an app that organizes receipts and warranties in one place, using AI to make it easy to track what you own and when coverage expires.

A Jira Cloud app that makes daily standups faster and sharper — an AI assistant that times the meeting, surfaces what changed since yesterday, and flags risks and blockers, reporting up to a 40% cut in daily meeting time.

An Amazon Alexa skill that answers any question using generative AI, with kid-safe guardrails on every response.

A WhatsApp bot that transcribes voice messages on demand — forward an audio to a specific number, get back a readable transcript instantly. Built solo, in my own time, with multi-language support so it works across different audiences.

A Trello analytics tool that turns boards into actionable metrics — idea to MVP in five months, grown to 500+ users, with a companion Alexa skill.