Engineering lead building production backends, iOS products, and AI-assisted developer tools.
I like owning the parts of software that quietly decide whether a product works: APIs, queues, deployment scripts, observability, data pipelines, Live Activity delivery, and the glue between mobile apps and backend systems.
- Engineering Lead at Arkade Club, owning backend infrastructure for Apple-featured apps used by 3M+ people.
- Creator of Pushboy, a Go-based SNS alternative for self-hosted push notifications, Live Activities, delivery receipts, scheduling, topics, and dispatch state.
- Primary contributor to App Store Connect CLI, with 31+ merged PRs across TestFlight, IAPs, subscriptions, pricing, localization, screenshots, auth, and workflows.
- Built and launched iOS products including Focus Rail, with Mapbox, SwiftUI, embedded SQLite, RevenueCat, and App Store workflows.
- Winner of the RevenueCat Shipaton 2025 $10,000 prize for Dripped, an AI wardrobe app with RAG-powered recommendations.
| Project | Why it matters |
|---|---|
| Pushboy | Go notification infrastructure with Postgres storage, worker pipelines, APNs/FCM dispatch, OpenAPI docs, Docker setup, and Live Activity support. |
| App Store Connect CLI | Open-source contributor to a 4k+ star CLI used to automate Apple developer workflows. |
| Focus Rail | Shipped iOS focus timer that turns Pomodoro sessions into train journeys with maps, history, subscriptions, and route data. |
| Dripped iOS | AI outfit recommendation product using Cloudflare Workers, TypeScript, embeddings, vector search, and LLM orchestration. |
| Pokemote | 70+ star MCP server and real-time TV control system using Server-Sent Events and semantic search. |
| ConvJobs | Hackathon-winning AI hiring platform with React/Vite, Convex, Clerk, OpenAI embeddings, and conversational candidate search. |
Go, TypeScript, Node.js, Swift, Python, PostgreSQL, MongoDB, SQLite, Docker, Cloudflare Workers, Azure, Loki, Grafana, Tempo, SwiftUI, Mapbox, RevenueCat, OpenAI, Claude, Codex, LangChain, RAG, vector databases, CI/CD, Linux VMs.
I use AI agents every day, but the bar is still production evidence: read the code, ship the narrow fix, run the tests, inspect the logs, and verify the real path.