Open Source Design Engineer

Arize AI

7d•$145,000 - $185,000•Remote

About The Position

We're looking for a Design Engineer to build ambitious interfaces at the frontier of AI development, evaluation, and observability. You'll design, prototype, and ship experiences that help AI teams understand, troubleshoot, and improve the systems powering their next-generation applications. Here's what makes this role different: Modern AI work is no longer purely human-driven. Arize Phoenix is a platform used by AI teams and AI agents such as Claude Code. You won't just craft great UX. You'll pioneer great AX (Agent Experience)—product surfaces that are delightful for people and interoperable, safe, and legible for agents.

Requirements

Have strong frontend engineering skills (React, TypeScript) paired with design sensibility
Think deeply about interaction design, information architecture, and developer experience
Are curious about AI systems and eager to learn how agents, evaluations, and observability work
Enjoy prototyping quickly and iterating toward polish
Care about accessibility, performance, and craft
Thrive in open-source communities and async collaboration

Responsibilities

Design and build compelling product experiences for Arize Phoenix across UI-driven and programmatic workflows
Create interactive visualizations for embeddings, traces, evaluation runs, and agent/tool workflows
Develop reusable components and templates for our design system, from concept to production
Build Agent Experience primitives: interaction contracts, human-in-the-loop workflows, and agent-friendly UI surfaces
Enable agents to query, navigate, trigger evaluations
Shape the patterns that define how humans and AI agents collaborate within a shared workspace
Prototype new AI features and UI patterns for visualizing performance, surfacing anomalies, and explaining AI behavior—especially for multi-step agent systems
Use AI-assisted UI generation to accelerate prototyping while maintaining craft, accessibility, and consistency
Push the boundaries of what's possible when your users include both humans and machines