AI Automation Engineer — Build & Hand Over Systems That Replace Manual Marketing Ops

We are hiring one automation engineer, and we are hiring selectively. Read the questions at the end before applying — they are the actual filter. We are an AI-first venture studio. We build and operate our own portfolio rather than servicing clients: consumer D2C brands and B2B SaaS products across several markets. Creative, funnels, tracking, media buying and the automation layer all sit in house. Here is the honest version of what we want. A traditional operation at our volume would need a marketing department of around forty people — analysts pulling reports, editors renaming files, coordinators chasing approvals, assistants moving data between six tools, someone manually reviewing every asset before it goes live. We do not want to hire those forty people. We want the systems that make most of them unnecessary, and we want the engineer who can see where those systems go before we ask. Two ways to work with us One — build with us ongoing. Hourly, long-term, 30+ hours a week, direct access to the founder and no client-services layer in between. You design and build the automation layer across the portfolio: creative production pipelines, media buying operations, voice and chat agents that handle real conversations, and internal systems that turn scattered messages into tracked work. Then the unglamorous half — monitoring, error handling, retries, alerting, and documentation good enough that it survives you being on holiday. Two — hand over what you have already built. This is a real part of what we are looking for, so do not skip it. If you already own working automations, we will buy them on a fixed-price basis: you hand over the build, we pay for it, you implement and document it in our environment. Bring your inventory. We are equally happy for you to take something strong from a public or open-source resource and implement it properly for us — we care that it works and that you understand it well enough to maintain it, not that you wrote every line from scratch. Tell us what you have, what it does, and what you want for it. What we need technically Not just no-code workflow tools. n8n is useful and we use it, but a large part of this role is server-side: standing things up on your own infrastructure, self-hosting, managing environments and secrets, keeping services alive and monitored. If your entire experience lives inside a hosted workflow builder's UI, this is not the right fit. And someone who writes code. We are explicit about this because we have hired the alternative. If your process is describing what you want to an AI chat window and shipping whatever comes back, you can produce something impressive that fails quietly three weeks later on security, scale, or an edge case nobody tested. We use AI heavily ourselves and we expect you to — but we need someone who can open the code, read it, and tell us why it will break. We are deliberately not listing our stack in detail. Show us what you use and why. Who this is for Someone genuinely obsessed with this field. Not "keeping up with AI" the way everyone writes on a CV. The kind of person who reads release notes and repos for fun, and who shows up on Monday saying there is a new model, a new node, a new approach, we should test this before our competitors do. In this field what worked last week is obsolete this week, and we need to hear it from you rather than tell it to you. Someone who understands the business, not just the tooling. Our previous automators built things nobody used. They waited for specifications instead of asking what the actual problem was, and quoted three days for work our non-technical founder finished in one. If you need a ticket before you can start, this will not work. Someone fast. Genuinely fast, not "I work efficiently" fast. Speed is the entire point of the role. How the hiring works We are AI-first, so our first interview is conducted by an AI voice agent, not a person. It asks technical follow-ups based on what you actually say, it checks claims against the web in real time, and it is deliberately hard. A human reads every transcript. Most people find it uncomfortable. If that already sounds like a reason not to apply, this is not your role. For the ongoing track we start with one paid scoped build rather than a long contract on day one. You tell us what access you need, you build one thing, we review the code. It goes both ways — you see how we work before committing. To apply Skip the cover letter template. Answer these four, briefly and specifically: Describe the most complex automation you have built. What did it do, how many runs a day did it handle, what did it cost to run per month, and what broke first? Which working automations do you already own and could hand over to us? List them with a one-line description and a fixed price for each. Confirm you have the right to transfer them. Describe something you have self-hosted and maintained yourself. What was the stack, where did it run, and what broke? Something you tested in the last month that we have probably not heard of — a model, a repo, a technique. One or two sentences on why it matters.