Core AI Engineer

You will own end-to-end pieces of the product. There is no queue to pull from and no PM to hand you stories. You work directly with the founders, identify what matters, build it, and watch it land.

Responsibilities:

• Own at least one named system inside the product, end to end. Context storage, retrieval, agent orchestration, or the integrations layer. Yours to architect, ship, and operate.
• Wrap LLMs into agent infrastructure that behaves reliably under real load, with the right tradeoffs across latency, cost, and quality.
• Ship daily. We treat the number of iterations per week as a primary metric.
• Define how a piece of the product earns trust, recovers from error, and degrades gracefully when the model is wrong.
• Build with AI as a primary collaborator. We use AI to build AI. We expect you to be fluent in the tooling and serious about the craft of using it well.
• Watch the work go live to real users and feel the responsibility that comes with that.

Requirements:

• You have shipped things people use. You have also deleted things you were proud of because a simpler version was better.
• You think of software as thinking made tangible. You write code that is easy to delete. You have internalized that complexity is a cost you pay every day, not a one-time decision.
• You understand AI systems deeply enough to reason about failure modes, context-window tradeoffs, latency vs. quality decisions, and when agent behavior is wrong vs. when the prompt is wrong. You have debugged things that do not have stack traces.
• You move fast but not recklessly. You know when a proof of concept is the right next step and when it is procrastination with a better name.
• You are uncomfortable with opacity. You add logs, not comments. You know the system you are building will be running while you sleep, and you build it like it will be.

Skills

AI, LLM, agent, agents, architect, architecture, artificial intelligence, large language model