About the RoleWe're hiring a Senior Applied AI Engineer to ship the AI features that designers actually use every day. You'll prototype, evaluate, and refine product capabilities - from prompts and agents to retrieval and tool orchestration - and turn raw model capability into reliable, delightful UX. This is a deeply hands-on role for someone who lives at the seam between AI capability and product, can navigate a large TypeScript codebase, and treats fast iteration over polished engineering as a feature, not a bug.
You'll own AI features end-to-end, capabilities like our AI Cursor, from prompt design and agent orchestration through to the React-aware code transformations that actually ship to users.
What You'll Do- Build user-facing AI features end-to-end: prompt design, agent flows, tool use, RAG pipelines and the production code that makes them work
- Own complex capabilities end-to-end. Take something like AI Cursor from spec to shipped: design the prompt and agent flow, build the retrieval, write the React code transformations, ship the UI, and own the eval
- Ship directly into our TypeScript monorepo - your code runs in production, not in a side service
- Read and reason about React code as data - the model's job is to modify React, and yours is to make sure it does so reliably
- Iterate quickly on capability vs. UX tradeoffs - when does the canvas need a fast path vs. a slow path, when should the model suggest vs. just execute
- Design product-facing evaluation: A/B tests, user-driven metrics, LLM-as-judge harnesses, behavioral regression suites
- Own the prompt-engineering, context-engineering, and retrieval layers for every AI feature
- Know when a capability has hit the limit of prompting/retrieval and needs fine-tuning or serving-side work - and raise it.
- Be the closest engineer to designers and product - translate user intent into AI capability
Must-Have Requirements- 8+ years software engineering experience, with demonstrable production AI feature work
- Production TypeScript experience in large, complex codebases - you've shipped non-trivial features in a TS monorepo and are comfortable owning code end-to-end in a JS/TS stack. This is a hands-on engineering role, not a prompt-tuning role.
- Strong prompt engineering, context engineering, and RAG fundamentals
- Experience designing agent flows with multi-step tool calling
- Hands-on experience with LLM APIs and orchestration frameworks (LangChain, LangGraph, DSPy, Vercel AI SDK, or equivalents)
- Ability to design and ship product-facing eval pipelines
- Strong product instincts - comfortable holding the pen on UX tradeoffs
- Move-fast disposition; comfortable with rapid iteration over polished engineering
Nice to Have- Ability to read and reason about React code - components, hooks, state, JSX as a tree. Our AI modifies React code, so understanding what it's producing is a major plus
- Any background in code-generation AI - Copilot-style completion, agentic code editing (Cursor, Cline, Aider, Devin), AST-aware transformations, code-specific RAG, code evaluation harnesses
- Experience with streaming AI UX (token streaming, partial renders, cancellation)
- Familiarity with WebSocket-based real-time systems
- Background shipping AI inside collaborative or canvas-based products
What You'll Build- Canvas-side AI capabilities, including AI Cursor and successor capabilities
- Multi-step agent flows that take a user goal and execute it end-to-end against a React codebase
- Retrieval pipelines over the user's design system and existing codebase
- Real-time, streaming AI experiences in the canvas
- The eval harnesses that tell us when a feature is shipping-ready vs. needs more work
Benefits- Salary: $250,000-$400,000 base salary
- Equity: Meaningful stock options
- Health Insurance: Best-in-class coverage for the employee and their entire family
- Location: San Francisco HQ
