Founding AI Engineer

Founding Engineer — Matchcard

Recruitment services | AI engineering | Building great tech products

Oslo (Mesh) · Start: asap

This is a fulltime role, but we can begin part time or evenings/weekends to clarify compatibility

I'm looking for one person to build Matchcard with me. Not a team, not an agency — one engineer who knows what production-grade software actually looks like, and who moves fast enough that we ship the next tool in weeks rather than quarters.

What Matchcard is, and where it's going

Right now: a set of small, sharp automation tools for IT consultancy and recruitment. Each one does a single job properly instead of being another module in a bloated ATS. Like micro recruitment services that are highly effective by themselves, and that can be put together to form a coherent end to end product.

Two are live:

· Competence Needs Clarifier — turns a client's messy, half-articulated need into a clean, exportable assignment description.

· CV Generator — a consultant uploads a CV or LinkedIn export and gets a polished PDF and

Word document, plus an optional video CV recorded in the browser or uploaded from a phone, on a shareable page.

Next in the queue: a requirement matcher that returns a ranked shortlist from the consultant bench, semantic search across a consultant pool in plain language, contract generation, client intel portal, competence hiring portal, and more.

The tools are the strategy, not the product. Each one earns its place standing alone, but they're built to fit together — converging into an end-to-end service for matching IT consultants to assignments. Consulting is where we start, not where the problem ends. We get there on the back of tools that already pay for themselves.

Where it comes from

I run Norway Consulting, an IT contract recruitment business — enterprise clients, consultants on assignment, twelve years of doing this the manual way. That business is profitable, it pays for this one, and it's the first customer and the first dataset. We build the tooling, prove it on our own operation, then take it out as a product.

What I care about most

You know what a well-built fullstack system looks and feels like. You've shipped software that real people depend on. You can stand up a proper development environment — sensible structure, CI, environments, secrets, deploys that don't require a ritual — and you do it because you've felt the cost of not doing it, not because a blog post told you to.

Backend depth, frontend focus. Most of the visible work is frontend, UX, and product. But it has to sit on a backend that someone with real depth designed — data models, API boundaries, auth, background work, the things that quietly decide whether a product can grow. I need both in one person, and the backend half is the one that's harder to fake.

Product sense. This matters as much as the code. You should be able to look at a half-formed idea and see the version that a recruiter would actually use on a Tuesday afternoon. Knowing which 20% to build, and which features would look impressive in a demo and get used twice — that judgement is most of the job.

Business outcome over output. The question is never "did we ship the feature." It's whether the thing we shipped made a placement faster, a proposal better, or a client stay.

Genuine excitement about AI engineering. Not as a checkbox — as a lever. You should be pushing AI tooling as far as it goes, both in how you work and in what the product can do. Claude, Codex, Cursor, agents, whatever gets you there. I work this way too, so it's not something we'll have to argue about.

The stack

· Next.js / React on Vercel

· A monorepo that also hosts independent Vite + React SPAs, served behind rewrites

· Relational database — currently Postgres. Postgres experience is an advantage, but comparable relational experience transfers fine.

· Object storage and video hosting for the video CV path

· LLM APIs for parsing, tailoring, translating, and matching

· Tailwind, Radix, framer-motion, GitHub Actions

All code and documentation is in English. Working language is Norwegian or English — take your pick.

Bonus, not required

· Norwegian. A real advantage — the clients, the consultants, and the contracts are Norwegian, even though we write code in English.

· Recruitment, staffing, HR tech, or marketplace experience

· Video and media infrastructure — hosting, codecs, browser recording, streaming. This part of the product has teeth and someone who's done it before would save us months.

· Infra you can run yourself: DNS, Docker, VPS, CI

· Design sensibility. Matchcard has to look like it was made by people who care.

How we work

Likely it will be a mix of onsite and remote due to a small team in the beginning. Fully onsite when the team is established and we make a collective push.

Mesh a few days a week, remote the rest — but not fully remote for the first stretch, because that's where the product gets decided and that goes badly over async text.

We write things down. Decisions get recorded with their trade-offs, so the next person doesn't re-derive them. Have a look at the docs in the repo — that's the standard.

Unusual ownership. Also unusual ambiguity. If you need a fully specified backlog to be productive, this will frustrate you. I come from frontend, UX, and product, so I'll have opinions about the interface, and I want someone who'll push back on them.

Terms

Salary / hourly rate: To be clarified

Equity: This is a freshly started initiative, and as such equity is an option once we know it’s a match

How to apply

It's OK if you just send me your CV and a few lines about what you are interested in, but preferably this:

1. Something you built and shipped. A link beats a CV. Tell me what was hard about it.

2. Half a page on this: given a consultant CV and a client requirement, how would you build a matcher that a recruiter would actually trust? I'm most interested in how you think about the failure cases.

I'll reply to everyone.