Senior Software Engineer, AI Growth

At Sanity, we're building the future of AI-powered Content Operations. Our AI Content Operating System gives teams the freedom to model, create, and automate content the way their business works, accelerating digital development and supercharging content operations efficiency. Companies like Linear, Figma, Riot Games, Anthropic, Spotify, Arc’teryx and Morningbrew are using Sanity to power and automate their content operations. In this role, you'll help build the full vertical for AI-native development at Sanity – from advancing our Model Context Protocol (MCP) server to shipping agentic systems that can scaffold Sanity apps, Studios, and full-stack frontends. You'll work closely with a designer, product manager, and fellow engineers to move fast, experiment constantly, and turn emerging AI capabilities into real products. If you're the kind of person who stays up late tinkering with new model releases and thinks in terms of what's possible rather than what's been done before, this is your role. Here's some of what we've shipped recently: - Sanity MCP Server https://mcp.sanity.io – our remote MCP server, now generally available https://www.sanity.io/blog/sanity-remote-mcp-server-is-generally-available and listed as one of eight official MCP servers in Vercel's v0. - Sanity Agent Toolkit https://github.com/sanity-io/agent-toolkit – agent rules exposed via MCP, plus a Claude Code plugin with skills and slash commands. - Auto-configured MCP in the CLI https://www.sanity.io/docs/changelog/beb93f6c-7913-4ea7-8bbb-847ba7b51b19– the Sanity CLI automatically detects code editors and configures the MCP server for you. WHAT YOU WILL DO - Advance our MCP server – add new capabilities, tools, and integrations that let AI agents interact deeply with Sanity's content platform. - Build agentic developer tools – ship systems that can generate Sanity apps and studios, as well as frontend code with minimal human guidance. - Design and run evals – build evaluation suites using Braintrust to catch regressions, measure improvements, and make data-driven decisions about prompt and model changes. - Build a Sanity MCP app https://blog.modelcontextprotocol.io/posts/2025-11-21-mcp-apps/ – create rich React interfaces on top of the MCP protocol for platforms like ChatGPT (using the new ChatGPT Apps SDK https://developers.openai.com/apps-sdk/build/chatgpt-ui) and Claude. - Move across the stack daily – one day you're designing a new MCP tool, the next you're polishing a React component or deploying a new microservice. - Apply prompt engineering craft – diagnose why models fail, steer them effectively, and build intuition for what makes prompts work at scale. - Ship production AI features – write clean, well-tested TypeScript and turn experiments into reliable capabilities that thousands of developers use. - Collaborate with the team – work closely with design, product, and engineering to iterate quickly and share what you learn. ABOUT YOU: - Based in Remote US or San Francisco / Bay Area – if you are in San Francisco/ Bay Area, you should be able to work at least 2 days a week from the office. - Strong TypeScript, React, and Node.js skills – you're comfortable moving between frontend and backend, and a pure specialist profile isn't what we're after. - Shipped AI-powered code to production – whether at work or through side projects, you've built real things with OpenAI, Anthropic, or Vercel AI SDKs. - Actively using MCP in your workflow – you've enhanced your own development with MCP servers. Bonus if you've built one yourself. - Fluent in the current AI landscape – you can articulate the differences between Claude Opus 4.5, Gemini 3, and GPT-5.2, and you have opinions on tools like Cursor, Claude Code, Lovable, and v0. - Hungry and curious – you're obsessed with the craft, experimenting constantly, and energized by how fast this space moves. - Strong prompt intuition – you can "decode" why a model is hallucinating or ignoring instructions without relying