Engineer at Anthropic, where I helped build Claude Code — the CLI that lets Claude work directly in your terminal. My work focuses on developer tooling, agentic infrastructure, and making AI assistants more useful for real engineering workflows. Day to day that means designing hook systems, building safety layers, and figuring out how to give Claude persistent context across sessions.

I studied computer science at Stanford University, where I got deep into systems programming, compilers, and human-computer interaction. That mix of low-level engineering and UX thinking ended up shaping a lot of how I approach developer tools — performance matters, but so does the experience of using the thing.

Before Anthropic I worked across a few different parts of the stack — backend services, infrastructure, and internal tooling — but I kept gravitating toward the intersection of AI and developer experience. Joining Anthropic and working on Claude Code was the natural next step.

Preview Ops — an AI-native infrastructure platform that turns a single prompt into a fully provisioned, deployed, and wired-up stack. Describe what you want in plain English, get a live URL in seconds. It also gives every PR a complete preview environment — frontend, backend, database, seeded data, SSL, custom subdomain — torn down automatically on merge.

The stack is TypeScript, Rust, and Go — a prompt engine that parses natural language into deterministic service graphs, a provider-agnostic provisioner layer, and an intelligence layer that predicts failures and auto-recovers.

I spend most of my spare time coding. I like picking apart developer tools, contributing to open-source projects, and prototyping ideas around AI-assisted workflows. When I'm not writing code I'm usually reading about programming language design or trying out new frameworks.

Follow what I'm working on: x.com/previewops