Senior AI Software Engineer, Agent Systems

👩🏻‍💻We build and ship agent platforms that do real work in production. This role is for an engineer who designs self-running agent loops and multi-agent swarms, not someone who tweaks prompts one turn at a time. You will own systems that find work, do it, verify it, and report back, with humans at the gates rather than in every iteration. 🧑🏻‍💻This is a hands-on, ship-first engineering role. You write production code other engineers trust, and you treat agents as software systems that need triggers, budgets, stop conditions, verification, and observability like any other. 📲What we mean by loop engineering: designing an agent system as a control loop with an explicit trigger, scope, action, budget, stop condition, and report, plus a verifier that can reject bad work, so it runs unattended and knows when to stop or hand back to a human. Not a bigger prompt. A system that prompts the agents for you. 🫵🏻What we mean by swarm agents: a coordinated fleet of specialized agents (planner, builder, reviewer, verifier, scout, coordinator) with file and task ownership, shared state, quality gates, and clean handoffs, run by an orchestrator rather than one do-everything assistant.

📊What we're looking for: -Strong engineering fundamentals. 5+ years writing production software that other engineers depend on. (Adjustable; we care more about what you have shipped than the number.) -Hands-on loop engineering. You have designed agent loops with explicit stop conditions, budgets, retries, and self-verification. You can explain the difference between a task on repeat and a real loop, and you know why the verifier matters as much as the maker. -Multi-agent or swarm experience. You have built or operated systems where multiple agents coordinate: orchestration, handoffs, shared state, ownership or locking, and quality gates. -Fluency with modern agent tooling. Claude Code or Codex style agents, sub-agents, persistent memory and skills files, tool and function calling, MCP, and reason-act-observe loop patterns. -Solid TypeScript and Node. Comfort with a service framework (NestJS or similar) and a typed data layer (Prisma or similar). -Cloud and delivery. AWS (ECS or Fargate or similar), Docker, and CI/CD. You can take something from repo to production yourself. -A verification mindset. You treat "done" as a claim to be proven, and you build the checks that prove it.