Product Engineer III

THE ROLE

We're looking for a Fullstack Product Engineer to join a small, fast-moving team building the next generation of AI-first music creation experiences. This is a fullstack, AI-native role where you'll own features end to end, from prototype through production, monitoring, and iteration. You'll work alongside engineers, PMs, and designers in a high-trust, high-autonomy environment, shipping work that real producers and music-makers rely on. Our team practices agentic engineering end-to-end, and we expect engineers to be fluent with tools like Claude Code, Cursor, and Codex to build production-grade features without sacrificing code quality or reliability.

WHAT YOU'LL DO

• Build and ship fullstack features across multiple AI-first products, including agentic music production and audio ML experiences like voice conversion, synthesis, and composition
• Work directly with PMs and designers to take rough ideas to shipped product quickly, owning the full arc from ticket to production
• Integrate and serve audio ML models in production and build the creator-facing surfaces that bring them to life
• Develop agents and AI-powered workflows that do real, meaningful work inside the product
• Use agentic coding tools (Claude Code, Cursor, Codex) as a core part of your daily development workflow
• Own features end to end, prototype, production, reliability monitoring, and iteration
• Uphold code quality and architectural standards even when moving fast, reading diffs line by line and writing tests for production-facing PRs
• Contribute to a codebase that improves over time, with integrated reliability monitoring and a measurable weekly impact on shipped features

JOB REQUIREMENTS

• Experience — startup / early-stage product experience where you owned significant scope and shipped real product
• Problem decomposition — You clarify scope, edge cases, and constraints before writing a single line of code
• AI fluency and control — You direct the model, write targeted prompts, reject unhelpful suggestions, and know when not to use AI at all