Shipped to real users
They can name the model, the traffic and the bill. Production is a different sport from a leaderboard.
Staging, not indexed. Highlighted text is a placeholder.
Everyone has an AI project on their CV now. We find the people who got one into production, kept it there, and can tell you what broke.
Usually when the demo works and the product doesn’t. You’ve proved the idea with an API and a prompt, and now you need evals, latency and cost under control. Rush it and you hire someone brilliant in a notebook who has never been paged at 2am.
They can name the model, the traffic and the bill. Production is a different sport from a leaderboard.
They measure quality before and after every change, and they built the test set themselves.
They know when to fine-tune, when to prompt, and when a smaller model does the job for a tenth of the price.
They ask what the user is trying to do before they ask which model to use.
A messy, realistic task: an LLM feature giving bad answers on a tight budget. They work through it live, and our anti-cheat checks the interview. Anything it flags becomes a follow-up question, never a verdict.
In the same interview we ask how they measured quality on something they shipped. People who did the work can go three questions deep. People who read about it can’t.
Employment history checked, and references confirmed through work email with people who worked with them on the system they claim.
Checked before you meet anyone. No surprises at offer stage.
Base salary at venture-backed startups. Draft figures, pending Jono’s sign-off.
| Level | London | Rest of UK |
|---|---|---|
| Junior (0-2 yrs) | £50k-£70k | £40k-£55k |
| Mid-level (2-4 yrs) | £75k-£100k | £60k-£80k |
| Senior (5+ yrs) | £100k-£140k | £80k-£110k |
| Staff / lead | £130k-£180k | £105k-£140k |
Our fee is fee and payment terms. With Pay as they stay you can spread it across your hire’s first months, so your runway stays yours.
You get 3-5 vetted candidates by a date we agree upfront, usually within 10 days of signing off the brief.
Some will, if the mission is real and the numbers are honest. We benchmark salary and equity with you before we start, so nobody wastes a final round on a package that was never going to land.
Sometimes, but most startups need applied engineers who ship. We’ll tell you straight which one your problem actually needs.