Supabase
AI Platform Engineer
About this role
Supabase is seeking a Staff AI Platform Engineer to design and build the core execution layer for an AI-native internal operating system. You'll develop an event-driven agent runtime with governance, evaluation, and safety systems built into the infrastructure itself, starting as the sole engineer on this foundational platform.
What you'll do
- Build production-ready agent execution platform with event queue, headless runtime, durable state, human review gates, and complete logging
- Design and implement evaluation layer with golden test suites, behavioral assertions, and CI gates to enforce safety before deployment
- Develop initial agent portfolio across executive, team, and individual contributor layers, plus meta-agents that observe and improve the platform
- Enforce governance through code architecture—risk tiers, least-privilege credentials, tool gates, and autonomy classes with no unsafe code paths
- Design human notification system with interruption budgets, message bundling, and value-first interaction patterns
- Create instrumentation to measure platform ROI, team maturity grades, and work absorption metrics from production data
What they're looking for
- Backend systems design and architecture
- Event-driven queue and workflow systems
- LLM orchestration and agent frameworks
- Evaluation and testing frameworks for AI systems
- Database design and SQL (Postgres preferred)
- Infrastructure and observability
- Governance and security controls
- Python or similar systems language
Opens the application — the Jobs AI extension fills it for you. Set up autofill
Opens the official application on the employer’s site. No login required.
Supabase
Supabase builds an open-source backend platform built on PostgreSQL, offering managed database services, authentication, and APIs to developers. The company is hiring infrastructure, database, and platform engineers to develop and maintain its core systems, including management APIs, deployment infrastructure, authentication services, and site reliability practices.
View all jobs at SupabaseLikely interview questions
- How would you design a runtime to be model-agnostic while ensuring consistent safety properties across different LLM providers?
- Walk us through your approach to building an evaluation suite that's trustworthy enough to gate production deployments—what would make you confident in its results?