The role

This is a Member of Technical Staff position for a full stack engineer who lives at the intersection of product engineering and applied AI. You will own features from the database to the browser, and you will ship AI agents that run in production against real traffic and real money.

This is not a research seat and it is not a prototype factory. We ship. You will be expected to architect systems, debug them when they break at 2am, and make them better the next morning.

What you'll do

• Design, build, and ship full-stack features end to end, from data model to UI, that go live and stay live.
• Build production-grade AI agents and the eval frameworks that prove they work.
• Stand up feedback loops and self-learning systems so the product gets sharper the more it runs.
• Treat observability as a first-class concern: logging, tracing, and alerting baked in from day one, not bolted on later.
• Own production incidents end to end, from the page to the post-mortem to the fix that stops it recurring.
• Handle large-scale datasets at the application layer without the system falling over.
• Use AI coding tools to multiply your output, while holding the bar on what good code actually looks like.

Mandatory tech stack

You should be fluent in most of this and able to ramp fast on the rest:

• Core: Node.js, TypeScript, NestJS, React, Terraform.
• Cloud: AWS, any combination of Lambda, ECS, RDS, S3, and the rest.
• AI: LLM and agent frameworks such as LangChain, LlamaIndex, or similar.
• Observability: Datadog, OpenTelemetry, CloudWatch, or equivalent.

What we expect you've done

• Built and deployed production-grade AI agents. Real systems with users, not demos.
• Architected and shipped full-stack features that are live in production today.
• Handled large-scale datasets at the application layer.
• Designed and implemented eval frameworks for AI systems.
• Built self-learning or feedback-loop systems.
• Made observability a first-class concern, with logging, tracing, and alerting in from the start.
• Managed real production incidents and system failures end to end.
• Used AI coding tools like Cursor, Claude Code, or Copilot to multiply your output, not as a crutch.
• Wrote production code before the LLM-assisted era, so you know what good code looks like without AI writing it for you.