Founding Infrastructure Engineer

Structured AI; New York (NY); On-site; USD 180,000 - 250,000; Full time

Benefits offered by this job

Health insurance

Dental insurance

Job summary

Structured AI is hiring its first infrastructure engineer in New York to own the execution platform for AI agents. You will ensure sandboxed, isolated execution with fast startup and controlled network egress, while preparing for scale as agent traffic grows.

You’ll also own the worker fleet, LLM traffic layer, and continuous deployment pipelines, shipping code in Python and FastAPI alongside product engineers. Join a founding-like team building the core of the product.

Qualifications

• Minimum 3 years running cloud infrastructure in production.
• Experience with containers and sandboxing for multi-tenant workloads.
• Production Postgres and Redis experience with on-call duties.
• Experience end-to-end with CI/CD and IaC.
• Familiarity with AI coding agents (Claude Code, Codex) used daily.

Responsibilities

• Own the execution platform for AI agents: sandboxed, isolated, scalable.
• Scale worker fleet from zero to hundreds of jobs; manage multi-tenant LLM traffic.
• Lead CI/CD, self-hosted runners, and incident response.
• Contribute in the codebase (Python, FastAPI) alongside product engineers.
• Define security and performance best practices as the system grows.

Skills

Cloud infrastructure

Production experience

Containers/MicroVMs

CI/CD pipelines

Postgres & Redis in prod

IaC & automation

AI coding agents usage

Job description

Structured AI builds AI agents that check construction drawings and CAD models. Design engineers at architecture and engineering firms run their drawing sets through us and get back verified findings: code issues, coordination clashes, things that don't match the schedule. A set can be 3,000 sheets.

We've raised $5m and are backed by Y Combinator, 20VC, Cherry Ventures, Airtree, and others. Syska Hennessy Group, GPD Group, and Auld & White already run their reviews through us.

ENR, Forbes, The Wall Street Journal, and Business Insider have covered what we're doing.

You'd be our first infrastructure hire on a team of ten engineers. Right now that means owning the platform the agents run on. Over time it means setting the technical direction for everything underneath the product: how we isolate untrusted code, how we scale, and what we run it on.

What you'd own

Our agents generate code and execute it as part of every check. You'd own the execution platform that runs it: sandboxed and isolated, with warm pools for instant startup and controlled network egress. That platform is the core of the role, and it has to stay fast and secure as agent traffic grows.

You'd Also Own

• The worker fleet: thousands of sheets per project, scaling from zero to hundreds of jobs and back.
• The LLM traffic layer: multi-tenant, metered, every token accounted for in customer billing.
• Streaming for chat and progress updates that survives deploys and dropped connections.
• The document pipeline: rendering, OCR, extraction on very large PDFs.
• CI/CD, the self-hosted runner fleet, and incident response.

You'd also ship in the codebase (Python, FastAPI) alongside the product engineers.

We're on Azure: Container Apps with KEDA-scaled jobs, Bicep, GitHub Actions, Postgres, Redis, Front Door, Entra ID. Models from Anthropic, Gemini, and OpenAI. We're hiring for depth in containers and isolation, not for any one tool. If you'd run part of it differently, you'll be the one deciding.

Who you are

• You've run cloud infrastructure in production for at least three years. Azure, AWS, or GCP; Azure is a plus, not a requirement.
• You've isolated workloads in production. Containers, microVMs, or Linux sandboxing in a system with real users. We'll go deep on this in the interview.
• Deep production container experience: Container Apps, ECS, AKS, EKS, or similar.
• You've owned infrastructure as code and CI/CD end to end.
• You've run Postgres and Redis in production and been on call for them.
• You use AI coding agents in your own work and check what they produce. We use Claude Code and Codex daily.
• You're in New York or willing to move here. We're in the office in the Financial District five days a week.

Strong signals

• Infrastructure for AI agents specifically: sandboxes for generated code, agent orchestration, inference fleets, vector search.
• KEDA or other event-driven autoscaling.
• Agent harnesses: MCP, tool-use loops, the Claude Agent SDK, LangGraph.
• Document processing, computer vision, or time spent in AEC.
• Python beyond scripts. Our deploy tooling is Python.

Compensation

• Base salary: $180,000 to $250,000, depending on experience.
• Equity up to 1%, sized for a founding engineer.
• Health and dental.