Software Engineer - AI Tooling Infrastructure

ABOUT SUPABASE Supabase is the Postgres development platform, built by developers for developers. We provide a complete backend solution including Database, Auth, Storage, Edge Functions, Realtime, and Vector Search. All services are deeply integrated and designed for growth. About the Role We’re looking for a Software Engineer, AI Tooling Infrastructure to join our AI Tooling team and help build and run the backend services behind Supabase’s AI products. The Supabase MCP server is one example of what the team runs in production, alongside the services that support agent skills and the dashboard assistant. You’ll work closely with the Platform, SRE, Auth, Billing and Edge Functions teams, and with the engineers on the AI Tooling team who build the product surfaces. The AI Tooling team runs its own services in production and carries its own on-call. Traffic from AI clients and agents behaves differently from traffic from people: hundreds of calls a minute, retry loops, and no reading of deprecation warnings. This role makes sure the services they use are reliable, correctly authorized and observable. It is a systems and infrastructure role rather than a product AI role, and it is not an SRE role. You will build and operate these systems, not only advise on them. This role is ideal for someone who thrives in async, fast-paced environments, owns what they ship in production, and is excited about building developer tools that scale to millions. WHAT YOU’LL BE RESPONSIBLE FOR In this role, you’ll: - Build and operate the production backend services behind Supabase’s AI products, including the MCP server: deployment, scaling, configuration and release. - Take ownership of how those services are structured, including splitting or reshaping a service when that makes it safer to run and easier to change. - Define, measure and meet reliability targets for the services the team runs. Set SLOs with the SRE team, track error budgets, and use them to decide what to fix first. - Participate in the AI Tooling team on-call rotation, and be a first responder for incidents in those services. Investigate, fix, coordinate with other teams during incidents, and write the follow-ups and runbooks. - Work the AI Tooling team support queue alongside on-call, and turn recurring issues into fixes rather than repeated manual answers. - Own authentication and authorization for AI and agent clients: OAuth apps, scoped tokens, and permissions that make sense for agents as well as humans, including how we change or deprecate a permission without breaking existing integrations. - Own observability for the services the team runs: metrics, traces, logs, dashboards and alerts, so we find problems before customers do. - Handle capacity and abuse for agent traffic: rate limits, cost and safety controls, and behavior under load or when a dependency degrades. - Improve the architecture of the services the team owns as they grow, including splitting or reshaping a service when that makes it safer to run and easier to change. - Work with the Platform team on shared code, review ownership and release process, and with the SRE team on infrastructure, alerting and incident practice. - Contribute to the code of the AI Tooling team’s tools, such as the MCP server, where it affects reliability and security, in partnership with the engineers who own the product surface. YOU MIGHT BE A GOOD FIT IF YOU Must have: - Have 5+ years of experience building and running backend services in production, and have carried on-call for them. - Write strong TypeScript or Go, and are comfortable working in the other. - Have designed and operated APIs with real consumers: versioning, error handling, rate limiting, and backwards-compatible change. - Know OAuth, OIDC, JWT and API authorization well enough to debug them in production and to explain where implementations commonly deviate from the spec. - Have a good understanding of PostgreSQL and relational databases. - Are comfortable with CI/CD, Docker and containerized deployments, and with infrastructure as code. Any tool is fine. - Can debug across services: follow a slow or failing request through several systems using metrics, traces and logs. - Communicate clearly in writing, work with a wide range of stakeholders, and thrive in async, globally distributed teams. - Are comfortable navigating ambiguity and moving quickly. Pick your centre of gravity. We expect you to be deep in at least one of these, and to tell us which: - APIs and auth: You have designed permission models, token scopes or OAuth flows for third-party or machine clients, and you have migrated them without breaking users. - Operations: You have run production services yourself: carried the pager, written alerts and runbooks, used SLOs or similar to decide what to fix, and made shared systems less risky over time. - Gateways and distributed systems: You have built or run proxies or gateways, or debugged hard failures across services (retries, partial failures, upstream outages, fail-open vs fail-closed decisions), and you can reason about load and degradation. Nice to have: - Experience with agent tooling (MCP or similar), or serving traffic from AI agents and LLM clients. - Experience moving a service out of a larger codebase while keeping existing clients working. - Experience with Grafana, OpenTelemetry, Logflare or a similar observability stack, and real SLO work. - Open-source contributions, especially to Node.js, Go, TypeScript or API tooling. WHAT WE OFFER - Fully Remote We hire globally. We believe you can do your best work from anywhere. There are no Supabase offices, but we provide a WeWork membership or co-working allowance you can use anywhere in the world. - ESOP Every team member receives ESOP (equity ownership) in the company. We want everyone to share in the upside of what we’re building together. - Tech Allowance Use this budget to set up your ideal work environment—laptop, monito