Agent Platform Engineer

About Kenn

Kenn is a venture-backed startup building next-generation systems to power the future of agentic work. Wes McKinney, Phillip Cloud, and Marius van Niekerk founded it in 2026. Before Kenn, Wes created pandas and co-created Apache Arrow.

kenn.io · github.com/kenn-io

The role

Kenn is hiring an engineer to own the internal systems that make our engineering team effective: the tools we build for ourselves, the agents that review, test, and maintain our code, and the machines and services they all run on. Today this work has no clear owner, and we fit it in around everything else. It matters too much for that.

You will work with everyone on the team. Within a year, our internal systems should support an agentic engineering team of ten people and run like a well-oiled machine: work is automated wherever it can be, everyone can see what is happening, and alerts fire only when a person needs to act.

What you will do

You will run the machines and services our engineers and their agents depend on, managed as code with OpenTofu and Ansible. The work includes:

• Set up development machines and sandboxes where engineers and their agents work, ready for a new engineer on their first day.
• Run the services the team relies on every day, such as automated code review and the machines that run our benchmarks.
• Keep every infrastructure change moving from a merged pull request to a verified deployment, with an incident opened when one fails.
• Keep the network access policy that is running the same as the one in the repository, and keep a tested way back into every machine when the network is down.
• Keep the CI runners fast and reliable enough that nobody thinks about them.
• Review the alerts regularly, and fix or delete any check that fires when nobody needs to act.

What you will build

Our internal tools are the team's entry point for everything: a web app, a CLI, and a Slack app, backed by an API that agents use as well. You will own their development, with coding agents doing most of the writing. The work includes:

• Monitoring that catches broken releases, missing downloads, failed installs, and CI failures on every platform we ship to, and opens an incident with the evidence attached.
• Benchmarks on every pull request and on the main branch, so regressions show up before they merge.
• Agents that grade pull requests, check whether a release is ready, investigate incidents, and maintain our repositories, running on worker machines you manage.
• Evals that tell us whether those agents are good at their jobs. When someone changes a prompt or swaps a model, we should know whether it helped.
• Daily roundups that tell each of us what needs attention and which pull requests are ready to merge.
• Ways for people and agents to ask for work where they already are, such as asking for a code change in Slack.
• Views that show anyone on the team what needs attention, with the same answers available to agents through the CLI.

These systems are how we scale our engineering process. As the team grows, so does what it needs from them, and you will own them the whole way.

Who we are looking for

• Builds with AI. You have built real systems with coding agents or on top of models, and you can show us. This is the one requirement we will not bend on.
• Knows evals. You have measured whether an AI system works on real tasks. You build evals from the failures you see in real runs, and you know when an LLM judge can be trusted.
• Builds tools for engineers. You have built internal tools, developer platforms, or CI and release systems that other engineers depend on.
• Runs infrastructure. You have operated Linux machines and services in production, managed as code with Terraform or OpenTofu and a configuration tool such as Ansible, Salt, or Chef.
• Ships with agents. You direct coding agents to build backend services and web frontends, and you can tell when what they built is wrong. Our tools are written in Go and TypeScript.
• Alert discipline. You know how to design monitoring that people trust, because it stays quiet until something is wrong.
• Design sense. You care how tools look and feel, and you build interfaces people want to use.
• Works across the team. You work with every engineer and turn their friction into fixes.

The level is open, and we expect the best fit to be senior. We care more about what you have built than where you studied or how many years you have worked.

Why join

You will build the platform an agentic engineering team runs on, at the point where the company's habits and systems are still being set. You will work with us on hard problems, and the tools you build will shape how the whole team works.

We are an early-stage company, so the role is broad and priorities shift week to week. Much of what lands on your desk is work that has no clear owner today. You become its owner, and each failure you turn into a check or a fix is one nobody has to handle again.