# Principal Software Engineer

**Company:** [Ardenir, Inc](http://jobs.workable.com/companies/sFCHEuVAj6Z7p32P94b5Mm.md)
**Location:** Seattle, United States
**Workplace:** hybrid
**Employment type:** Full-time

[Apply for this job](http://jobs.workable.com/view/ecdb6708-3285-41e6-acdc-61a363859fd5)

## Description

### About Ardenir

Ardenir is building the AI ad agency of the future. We run paid acquisition end-to-end across all the major ad platforms using AI agents that plan campaigns, produce creative, manage spend, and attribute value.

### The Role

As Ardenir's Principal Engineer, you'll own how we build software. We've adopted an AI-forward approach in which agents write our code. When a person has to review a line by hand, we treat it as a defect; a sign that something failed in a specification, a test, or a safeguard. Your job is to make sure our AI practices produce code that is correct, reliable, and gets better over time, and to personally take on our most difficult technical problems.

You'll report to the CTO, who builds and leads the engineering team. Between you, you'll hold the deepest technical responsibility in the company. The CTO owns the overall architecture, the team, and the company's technical direction, while you own how code gets written and verified.

We care about the quality of your work and the clarity of your thinking, not your years of experience or the size of any team you've led. You are an engineer others turn to when they're stuck, someone who has already reshaped how a team builds software alongside AI agents. In a few years, the systems you shape will manage hundreds of millions of dollars in ad spend across thousands of customer accounts.

### What You'll Own

**The development approach.** Our specifications, and the workflow that carries a task from plan to finished, verified code. We treat the specification as authoritative and generate the code from it. You judge when a task needs a full specification and when a brief instruction will do.

**The intelligence in the product.** The reasoning that makes Ardenir's agents effective, built into our business-logic layer, including custom models where they give us an edge, for example in attribution.

**Verification.** The automated checks that confirm the agents' code matches the specification, fits the architecture, and stays within its safeguards — plus separate agents that review the work rather than letting the author judge itself.

**Coordinating the coding agents**. How work is divided among them, how they pass tasks between one another, and how we catch the plausible-looking mistakes these models produce.

**The most demanding technical work, hands-on.** The core of the agent system and our ad platform integrations, such as Meta, Google, TikTok, and Amazon — including their rate limits, authentication, failure modes, and constant changes.

**Reliability and containment.** Catching problems before they reach a customer's account, and limiting the damage when something goes wrong with live spending.

**The standard for quality.** The patterns, conventions, and definition of finished work — written down so agents follow them and no one enforces them by hand.

### What You Won't Own

The CTO owns the overall technical vision, hiring and leading the engineering team, security and compliance, and the company's external technical decisions. The Head of Performance owns marketing outcomes and defines how the system should behave to achieve those outcomes. The CEO owns sales, fundraising, product strategy, and the financials.

## Requirements

-   You've delivered production software under real deadlines, making sensible scope and trade-off decisions to ship rather than gold-plating.
-   You've built distributed systems that stay correct and resilient under load and failure, where mistakes have real consequences.
-   You've built and run agent-based systems in production — agents doing real work, operating reliably, not prototypes or experiments.
-   You've built the frameworks that let agents write software — the development machinery itself, not just worked within someone else's.
-   You can evaluate and improve agent behavior in production: eval design, tracing and observability, and turning that signal into measurable gains in reliability.
-   Full-stack, in earnest: backend, frontend, and data infrastructure.
-   AWS, or comparable cloud experience.

Preferable:

-   Experience with our stack: Rust, TypeScript, AWS-native.
-   Experience with ad-platform APIs, such as Meta and Google.
-   A background in adtech, martech, or performance-marketing infrastructure.
-   Hands-on work with multimodal generative models in production.
-   Experience taking infrastructure through SOC 2 and GDPR from the start.

## Benefits

-   Unlimited PTO
-   Health Benefits
-   401K