AI Platform Engineer
Remote • Remote, Global • FullTime
Posted 2mo ago
Job Location
Remote, Global
Tech Stack
Remote Work Policy
Fully remote
Employment Type
FullTime
Categories
AI Agent Engineer
About the job
Supabase is seeking an AI Platform Engineer to develop the execution layer for its internal AI systems. The company is building an AI-native internal operating system where AI handles significant operational tasks. This role involves creating the platform that runs agents, including an event-triggered queue, a model-agnostic runtime, durable state management, human review gates, atomic rollback, and comprehensive logging. You will also build the evaluation layer to ensure trustworthiness and develop agents for various functions, from executive reporting to platform monitoring and improvement. The role emphasizes a governance-heavy environment, with agents risk-tiered and subject to review depth, evaluation requirements, and human approval. The engineer will be solely responsible for closing architectural decisions, owning the infrastructure end-to-end, and instrumenting the system for performance reporting.
Responsibilities
- Ship the agent platform to production, including an event-triggered queue, headless model-agnostic runtime, durable state, human review gate, atomic rollback, and complete run logging.
- Own and activate the evaluation layer, implementing golden suites with behavioral assertions, judge criteria with rubrics, safety cases, and a CI gate.
- Build and register the agent portfolio, covering reporting, drafting, linting, triage, and question-answering agents, as well as a meta-layer for platform observation and improvement.
- Enforce governance in code through risk-tiered pipelines, least-privilege credentials, tool-permission gates, decision audit logs, and autonomy classes.
- Design the system's interaction with people, including interruption budgets, message bundling, and user-friendly information delivery.
- Own the platform tooling, including the compiler, validator, inventory integrity, and distribution paths for context and capabilities.
- Compute operating measures from production data to create a maturity grade per team.
- Instrument the platform's return by logging agent work and computing monthly figures to quantify system value.
Requirements
- Demonstrate recursive thinking by building generators and compilers rather than individual artifacts.
- Apply inversion thinking by starting from failure scenarios to inform design.
- Utilize AI-native execution by employing agentic tools on real work.
- Close open architecture decisions.
- Own the infrastructure end-to-end.
- Instrument the system to report its own return.