About the role
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 to join our AI Tooling team and help shape how developers and agents build with Supabase. You'll build and own our AI tooling, including agents, the MCP server, agent skills and the dashboard assistant, and as the space moves you'll build the tooling for new technologies too. You'll build it all eval-first: every tool ships with evals and metrics, so we know, with evidence, that it makes agents better rather than worse.
You'll work closely with the Docs, CLI, Frontend and SDK teams, and with the AI Tooling Infrastructure engineers who run these services in production. Other teams contribute to AI Tooling to improve Supabase's agentic capabilities, and you'll help them do that while keeping the quality bar high.
This role is ideal for someone who thrives in async, fast-paced environments and is excited about building developer tools that scale to millions.
WHAT YOU’LL BE RESPONSIBLE FOR
- Build and improve the AI tooling we run today, including the MCP server, agent skills, agents and the Assistant.
- Make new Supabase products usable by agents from day one: tools, skills and evals for new product surfaces. Also build tooling for new AI technologies as they emerge: prototype quickly, evaluate honestly, and decide with the team what is worth taking to production.
- Drive an eval-first approach: build and improve evals, instrumentation and feedback loops, measure tools against a no-tool baseline, and help define the quality KPIs and gates that stop regressions shipping.
- Understand how customers and agents actually use what we ship, and turn that into improvements to our tooling, interfaces and documentation.
- Design how documentation and product knowledge are exposed to agents, and measure whether it works.
- Build patterns and tools that help customers add AI to their own apps, such as MCP on Edge Functions, vector embeddings and agents running on Edge Functions.
- Enable teams to improve Supabase's agentic capabilities through contributions to AI Tooling: clear contribution paths, templates, documentation and review practice.
- Centralise shared AI logic so teams reuse it rather than duplicate it.
- Ship tools that are battle-tested, with strong tests, benchmarking and stress testing where it counts.