Software Engineer, AI Agent Governance (Contract)

About CHURPRIMO

AI agents are starting to act on people’s behalf. The hard question isn’t whether they’re fast or busy; it’s whether they’re still doing what they were trusted to do. CHURPRIMO is building runtime governance for AI agents: a system that watches how agents behave as they work, recognises when they drift from what’s legitimate, and supports the right response.

We’re a small, close team with a working system already running across a fleet of agents. We don’t do titles. People are here because they choose to be.

The role

We’re looking for a dedicated engineer to take a running system and make it trustworthy: review and evolve the code, measure whether it really works, and help it grow into its next version. You’ll work closely with our founder, whose thinking the whole system comes from, and with a small team who will give you context, room and support.

What you’ll work on

• The scoring engine. A Python service that turns agent behaviour into a coherence score in real time. You’ll redesign parts of it, and prove each change with the test suite before it ships.
• Evaluation. Designing tests that show whether the score detects what it claims to, and being honest about where it doesn’t yet.
• Code review and quality. The codebase was built with a professional, agent-assisted workflow: an engineer working with AI coding agents, every change through pull requests and review. You’ll carry that forward, set standards and own what gets merged.
• A small, self-hosted fleet. Docker-based services on our own hardware: making them reproducible, observable and easy to rebuild.
• Making the system visible. Short weekly updates and regular walkthroughs, so the team can see the system run and knows what’s working, what isn’t and what’s next. Things breaking is normal here; saying so early is what matters.

Who we’re looking for

We care more about how you think than how many years you’ve been doing it. Seniority isn’t the filter, but solid engineering is: you’ll need to show real work you’ve built or improved, and the mindset to grow with something new.

You’ll likely have:

• Strong Python, and comfort reading and improving code you didn’t write
• A habit of testing whether things actually work, not just whether they run
• Experience with Docker and running services day to day
• Experience using AI coding agents in a disciplined, reviewed workflow
• The ability to explain technical things clearly to people who aren’t engineers

It’s a bonus if you have:

• Experience with embeddings, similarity measures or evaluation of ML systems
• TypeScript
• An interest in AI agents, safety or governance

Above all, you’re:

• Curious, kind and comfortable with ambiguity
• Open about problems, and quick to say when something isn’t working
• Excited by ideas that are new and still taking shape

Why you might love this

• You’ll shape something genuinely new from close to the start, not maintain someone else’s finished product
• Real ownership: your work goes straight into a system that’s already running
• A small, kind team that values honesty over polish
• Work directly with the founder whose ideas the system is built on

The details

How we hire

1. 01A short conversation about you and how you think
2. 02A short paid trial of 15 hours over one to two weeks, on real, contained work and including time with our founder, paid at a flat NZ$140 an hour
3. 03A decision together and, if successful, straight into a structured first month

How to apply

Send your CV, a link to your GitHub (or GitLab, or a portfolio), and a few lines on why this caught your attention, to people@churprimo.org. Point us to one or two things you’re proud of: code you’ve built, a pull request you’ve reviewed, or something you’ve written. If your best work is private, a short description of what you did and why it mattered is fine.