Senior Full-Stack Engineer • Based in Israel 🇮🇱
I build AI-native developer tools, full-stack products, and mobile apps - from architecture to production.
I write a lot of TypeScript and just enough Swift to be dangerous.
- 🚀 I build AI-native developer tools, full-stack products, and mobile apps — from zero to production.
- 🎨 Frontend applications and platforms built with React, Next.js, React Native, TypeScript, and production-grade UI architecture.
- 🧱 Backend systems with Node.js/Hono/TypeScript, NestJS, Python/FastAPI, Go, PostgreSQL/SQLite, Redis, GraphQL, and service/API architecture.
- 📱 Mobile apps across React Native and native ecosystems — including Swift-based tooling and experiments.
- 🛠️ I have a soft spot for Developer Experience (DevEx) — from AI coding workflows and internal tooling to CI/CD, release automation, docs, and smoother pull-request flows.
- 🔭 Currently building: Threadbase — my personal AI-native development environment for browsing, searching, starting, and resuming Claude Code & Codex sessions across devices.
- 🐦 Wrote Gitty — idiomatic Swift bindings for libgit2, for apps that don't just read Git, they write it.
- 🌱 Always poking at something new — if it compiles on the first try, I get suspicious.
- 💬 Ask me about TypeScript, frontend architecture, developer tooling, AI coding workflows, mobile development, or building products end-to-end.
🛜 I'd tell you a UDP joke, but you might not get it.
AI is deeply integrated into how I design, build, and ship software.
I continuously refine my engineering environment around Claude Code, Codex, Cursor, MCP servers, custom tooling, multi-agent workflows, and local automation. Rather than relying solely on off-the-shelf tooling, I enjoy building my own infrastructure, workflows, and developer tools that make AI coding agents feel like teammates instead of assistants.
Examples:
- Threadbase — a cross-platform AI-native development environment, originally built to solve my own workflow and later open-sourced for others. It lets me browse, search, start, resume, and monitor Claude Code & Codex sessions across devices.
- tmux-ai-necromancer — recover AI agent sessions after crashes and reboots.
- gh-mcp-proxy — extends GitHub's official MCP server with custom REST capabilities.
- docs-tracker-mcp and google-play-mcp-bridge — MCP tooling for documentation and Google Play workflows.
- ccpm, claude-marketplace, npm-package-owner, npm-registry-manager, and ralph-loop-file — utilities around AI-assisted development, package management, and agent workflows.
- A private Agentic OS / Second Brain built around Obsidian, MCP servers, a structured knowledge graph, an append-only agent run ledger, and a React-based operations dashboard.
💬 Languages
🎨 Frontend & Mobile
🧱 Backend & Data
🏗️ Testing, DevOps & Tooling
Beyond product development, I enjoy building the engineering systems that support it: CI/CD, release automation, deployment pipelines, and quality gates.
- GitHub Actions pipelines for build, test, release, and deployment.
- Release automation with semantic-release, conventional commits, changelogs, and Slack notifications.
- CI/CD with linting, type checking, E2E testing, preview deployments, and quality gates.
- Deployments to platforms like Vercel and Netlify, backed by clear documentation.
Representative examples: AutoKitteh WebPlatform, one-more-highlight, Gitty, and GittyKit docs.
A few principles I try to consistently follow:
- Code quality: strict linting, clear conventions, and maintainable architecture.
- Developer experience: Storybook, documentation, reusable components, and helpful tooling.
- Reliability: automated testing, CI quality gates, and regular dependency updates.
- Communication: meaningful PRs, conventional commits, and clear documentation.
- Think first: understand the problem before implementing the solution.
| Project | What it does |
|---|---|
| Threadbase | Cross-platform AI-native development environment for browsing, searching, starting, and resuming Claude Code & Codex sessions. Clients: mobile · streamer · electron · vscode · intellij · landing |
| Gitty | Idiomatic Swift bindings for libgit2, for apps that don't just read Git, they write it. |
| tmux-ai-necromancer | tmux-resurrect for your AI agents — bring Claude Code & Codex sessions back after tmux crashes and reboots. |
| gh-mcp-proxy | MCP proxy fronting GitHub's official MCP server, adding custom GitHub REST tools. |
| autokitteh/web-platform | Frontend platform I led as Engineering Lead — simplifying the automation of distributed systems. |
| ali-sum | Web app + Chrome extension that scans AliExpress order history into spending analytics. |
| groceries-bot | Telegram-controlled shared groceries list with a Next.js web UI, one service layer driving both the bot and the app. |
| yad2-scanner | Headless apartment/rental-listing scanner with Telegram + Gmail alerts, Neon Postgres, and per-user schedules. |
| one-more-highlight | React + TypeScript text-highlighting library for search matches and rich text rendering. |
Beyond developer tools, I also enjoy building small, practical automations that solve everyday problems:
- groceries-bot — a shared Telegram groceries bot with a web UI, built around one reusable service layer.
- yad2-scanner — apartment/rental scanning automation with Telegram and Gmail notifications.
- ali-sum — AliExpress order-history analytics via a web app and Chrome extension.
There are 10 kinds of developers: those who read profile READMEs to the end, and those who don't. Thanks for being the first kind. 🍻