2 months ago
Remote, WorldwideSenior
Responsibilities
- Ship and operate an event-triggered agent platform with a model-agnostic runtime, durable state, human review, atomic rollback, and complete run logging.
- Build evaluation infrastructure with golden suites, behavioral assertions, judge rubrics, safety cases, pass thresholds, and CI regression gates.
- Build and register reporting, drafting, linting, triage, question-answering, and meta-level agents.
- Enforce risk tiers, least-privilege credentials, tool-permission gates, decision audit logs, ownership requirements, and human approval controls in code.
- Design communication systems with interruption budgets, message bundling, and useful context before requests.
- Own the compiler, validator, inventory integrity, and distribution paths for context and capabilities.
- Compute team maturity measures from production data and create a ledger measuring work absorbed and platform return.
Requirements
- Have shipped production LLM agent systems with real users, operational history, and incidents you can discuss.
- Design formal evaluations using golden sets, behavioral assertions, judge rubrics, pass thresholds, and CI gating rather than relying on spot checks.
- Have deep API experience with systems where work actually lives and have authored MCP servers.
- Own infrastructure end to end using Python on GCP, including provisioning, deployment, monitoring, rollback, cloud warehousing, and infrastructure as code.
- Demonstrate strong judgment about how software contacts people and how notifications affect trust.
- Preferred qualifications include public work such as an open-source agent framework, MCP server, evaluation harness, or cited writing on agent reliability.
- Preferred qualifications include LLM observability, agent-run tracing, per-run cost attribution, and experience building an internal platform adopted voluntarily by non-engineers.
Benefits
- Fully remote, globally distributed work with no offices and a WeWork membership or coworking allowance.
- ESOP equity ownership for every team member.
- Tech allowance for equipment such as a laptop, monitor, or headphones.
- Supabase covers 100% of employee health insurance and 80% for dependents wherever available.
- Annual company off-sites in a new city.
- Flexible asynchronous work schedule.
- Annual professional-development allowance for courses, books, conferences, or other learning.
Tech Stack
Categories
About Supabase
Supabase builds an open-source Postgres-based backend platform for developers, bundling database, authentication, storage, realtime APIs, edge functions, and vector search. It monetizes through a managed cloud service and enterprise offerings while remaining deployable self-hosted. Founded in 2020 and privately held, the company operates globally with a fully remote team. Developers use Supabase as a Firebase alternative to launch quickly and scale production applications on PostgreSQL.