Lead Full-Stack AI Engineer

Why this role exists

We're building AdoptionAgent - the next evolution of Userflow's in-product Adoption Agent alongside AI features inside the Userflow builder. These two surfaces are how we believe the product adoption category gets rebuilt over the next 24 months: from authored flows toward agentic experiences that adapt to each end-user in real time.

What you'll own

• AdoptionAgent, end-to-end: the full development cycle of the agentic experience that customers' end-users interact with inside their own apps. This spans the UI, the agent orchestration, the structured map of each customer's app that gives the agent the knowledge it needs to act, and everything in between.
• High-stakes surface area: it's the user-facing AI.
• Userflow's AI-native engineering workflow: prompt patterns, agent UX heuristics, eval pipelines, internal tooling that makes AI-augmented work the default for both FE and BE engineers.
• Userflow's AI-native engineering workflow: defining how the org adopts AI in CI/CD, code review, and IDE-native development.

What we need you to bring

• Hands-on AI feature shipping → you've personally built and shipped LLM-powered product features. You can talk about prompt iteration, eval methodology, latency/cost tradeoffs, and when not to use an LLM.
• AI-native development workflow → you ship in Cursor, Claude Code, Windsurf, or equivalent. You've set up AI-augmented CI/CD pipelines and AI code review (CodeRabbit, Greptile, Claude review, internal equivalents). You can articulate what AI tooling has actually changed about your work, not just that you use it.
• Track record of raising the bar around you → you've changed how a team or org ships, not just how you personally ship.
• Comfort working across the stack. You may be deeper on one side than the other (see below), but you can read, reason about, and contribute meaningfully to code on either side.
• Frontend specialization: Deep experience building DOM-based JavaScript SDKs code that gets embedded into other companies' apps. You've shipped against MutationObservers, shadow DOM, CSS isolation, cross-origin constraints, and the performance realities of being a passenger script.
• React + TypeScript fluency at depth from state and rendering to where complexity should live.
• Backend specialization: Production Elixir experience at depth → operational reality of running Elixir services that real users depend on.
• At-scale WebSocket experience at significant concurrent connection counts. You understand connection lifecycle, presence, backpressure, and message delivery guarantees under real load.
• 5-8+ years building user-facing front-end systems, with at least 3 at the Senior+/Staff level shipping product surfaces that real customers depended on.

Nice to have

• Experience with agentic systems (multi-step tool use, planning loops, agent UX patterns)
• Experience with eval pipelines for LLM features in production
• Visual builder / canvas / drag-drop product experience — Userflow's core IP is a no-code visual flow builder, so background in canvas/drag-drop products (Figma, Notion, Webflow, Framer) translates directly
• Open-source contributions to AI tooling, agent frameworks, or DX tools