Staff Platform Engineer

Role Overview

Staff Platform Engineer (Infrastructure) at Postscript, helping build and operate Acorn—Postscript’s next-generation CDP and messaging platform. The platform consists of ~52 Rust/Python services running on EKS and is designed to handle 500K+ events/sec for 20K+ merchants, with infrastructure operations automated via AI-executed, self-verifying runbooks.

Primary Duties

• System & Platform Design: Own infrastructure topology, scaling and blast-radius boundaries, build/deploy graph, database ownership boundaries, and cost ceilings—decisions the agents execute but cannot make.
• Guardrail Engineering: Create validation gates, idempotency checks, drift detectors, and self-verifying runbooks to prevent damage by both humans and agents.
• Escalation Tier: Diagnose novel failures (e.g., consumer lag, DLQs, materialized-view chains, consent ordering) and author the initial runbook for agents to follow.
• Reliability, Capacity & Cost: Own the 500K events/sec target through load testing, Karpenter/HPA tuning, database sizing, SLOs, and alerting.
• Security & Secrets Boundary: Own IAM/IRSA, secret rotation, the auth chain, and define what the agent/MCP surface is allowed to do.
• Agent-Fleet Operability: Keep runbooks, skills, and agent context synchronized with real-world state to avoid executing stale/incorrect procedures.
• Incident Leadership: Lead incidents and convert every fix into a new runbook plus regression test.

Requirements

• 5+ years operating production Kubernetes and AWS cloud infrastructure at scale.
• Terraform and GitOps/Kustomize discipline; strong bias toward idempotent, automated, self-verifying operations.
• Incident response experience with distributed systems failures (databases, streaming, consumer lag, data integrity).
• Data systems fluency: how Postgres, a columnar store, and a log/streaming bus behave under load.
• Ability to work alongside AI agents by writing guardrails, runbooks, and checks that keep automation safe; know what must remain human-in-the-loop.
• Strong instincts to “encode as a check,” remove manual steps, and treat incorrect runbooks as production bugs.

Nice-to-Haves / Emphasis

• Operational reliability mindset and systems thinking focused on preventing harm rather than just performing deployments.
• Comfort owning novel failure modes and turning them into durable automated procedures.