Staff Product Designer - Agent Experience

Design the full programmatic interface, not just the CLI. The agent operates Render through many doors — the CLI, the API, MCP, SDKs, Blueprints, and Skills — and they should feel like one coherent product, not a set of independently evolved tools. You'll own the interaction model and the error language across all of them, so an agent can run deploy → verify → debug → iterate without a human in the loop. A one-command path from empty directory to running app matters, but so does an API call an agent can reason about and an MCP surface it can operate reliably. An error message an agent can't parse and recover from is a broken feature, and you'll treat it that way — on every surface.

Design documentation as an agent-facing product. Docs aren't just for humans anymore; they're how an agent learns to operate the platform. You'll design Blueprints, Skills, and reference material that teach an agent to use Render correctly — treating docs-as-experience as a first-class programmatic surface, with the same rigor as any command or endpoint.

Close the human-operated gap in onboarding. Today the default first-deploy path runs service-by-service through the UI — the one thing an agent fundamentally can't operate. You'll design a configuration-first path where the whole system is described in code (Blueprints) and deployed in one motion, so the programmatic loop doesn't break at step one.

Design the dashboard surfaces where agent work becomes visible to humans. When an agent deploys, retries, or fails on someone's behalf, a person needs to see what happened, why, and whether to trust it. You'll design the review, supervision, and activity surfaces for agent-driven work — web UI held to the same rigor as any flagship flow, because this is where confidence in the whole agent story is won or lost.

Design agent auth and access. Consent, scoping, and token management — so a developer can grant an agent or third-party app exactly the access it needs, see what that access is being used for, and revoke it without fear. Least-privilege should feel legible and safe rather than all-or-nothing. This is half policy design, half interface design, and it's where "can I trust this?" gets answered.

Design one model, expressed everywhere. The core objects of the platform — applications, services, environments — should mean the same thing whether you meet them in the dashboard, the CLI, the API, an SDK, MCP, or a Blueprint. You'll design that model once and hold every surface to it, so a human and an agent are always looking at the same product.

Surface the agent-operable paths at moments of intent. Copy-a-prompt CTAs, post-signup CLI install, post-error tooling suggestions. The programmatic paths already exist but aren't discovered — you'll place them where developer intent is highest.

7+ years of product design experience, ideally on developer tools, infrastructure, APIs, or another technical, credibility-driven product. You may have worked as a product designer, a designer who codes, or something harder to label — the title matters less than the work. At this level, we expect someone who has owned a major surface or product area end to end and set the direction others built against.

Craft that spans GUI and non-GUI surfaces. You have an eye for detail that borders on obsessive — the wording of an error, the shape of a command, the structure of an API response, the spacing of a dashboard — and you're as rigorous about a CLI message or an SDK method as you are about a screen. Your portfolio shows strong web UI work alongside text-first thinking: command design, API ergonomics, MCP/SDK design, docs-as-experience, or conversational interfaces.

You design the whole journey, not a surface at a time. Your portfolio shows end-to-end experiences — a user carried coherently across interface boundaries, CLI to API to SDK to dashboard and back — where you designed the model once and expressed it everywhere it appears, rather than polishing one surface in isolation.

A genuine point of view on what AI-native products mean for design — including how a product should show up to an agent, and not only to a person.

Comfort near code. You can read it, prototype with it, and collaborate with engineers without a translation layer in between.

The judgment to define quality where no established playbook exists, and the rigor to hold that bar once you've set it.

Clear, direct communication and cross-org influence. You can make the case for AX to technical and non-technical partners alike, in rooms where it's easy to treat it as purely an engineering concern — and you can align product, engineering, and leadership around a direction without formal authority over them.

Experience designing or shipping a CLI, SDK, API, or MCP surface used at scale.

Experience designing agent-facing documentation, Blueprints, Skills, or other material intended to teach a model how to operate a system.

Familiarity with OAuth, scoped tokens, or least-privilege access models — or an appetite to make permissions feel humane.

Hands-on experience building with coding agents or LLM-driven tooling, with a view on where the agent loop breaks down today.

Design-engineering literacy: motion, prototyping, or front-end fluency.

Experience evolving a developer-first product toward larger teams and enterprises without losing the original audience.

Writing, talks, or side projects that show your craft and point of view on developer experience.

This is a new category, and nobody has all the answers yet — including us. If you're the kind of designer who's curious enough to figure things out, ship, learn what's wrong, and try again, we'd love to hear from you.

If this role excites you but you don’t meet every single requirement, we’d still love to hear from you—your unique experience might be just what we need.